Discussion:
Gap in graphs
sanane enenand
2012-10-09 19:08:48 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/68968#68968

--------------------------------------------------------------
hi everyone,

i have the next problem... gap in graphs....

i have changed the version snmp of devices in the 'Configuration Properties' from v1 to v2c, because in the forum there where a few with the same problem and this solved them out.
but eventually i have STILL the gaps :(


i got also sometime have errors/warnings in my localhost:

threshold of high event queue exceeded: current value 5000.00
threshold of zenperfsnmp cycle time exceeded: current value 115.19



this is from the log of my deamons :

zenstatus :

2012-10-09 20:58:34,146 WARNING zen.ZenTcpClient: IP Service http is down
2012-10-09 20:58:34,147 INFO zen.ZenStatus: Finished 1 jobs (1 good, 0 bad) in 15.00 seconds
2012-10-09 20:59:41,314 WARNING zen.ZenTcpClient: IP Service http is down
2012-10-09 20:59:41,315 INFO zen.ZenStatus: Finished 1 jobs (1 good, 0 bad) in 15.00 seconds

zentrap:

2012-10-09 20:59:01,967 ERROR zen.zentrap: Discarding oldest 24 events because maxqueuelen was exceeded: 5024/5000
2012-10-09 20:59:11,968 ERROR zen.zentrap: Discarding oldest 22 events because maxqueuelen was exceeded: 5022/5000
2012-10-09 20:59:21,969 ERROR zen.zentrap: Discarding oldest 20 events because maxqueuelen was exceeded: 5020/5000
2012-10-09 20:59:31,969 ERROR zen.zentrap: Discarding oldest 16 events because maxqueuelen was exceeded: 5016/5000
2012-10-09 20:59:36,970 ERROR zen.zentrap: Discarding oldest 38 events because maxqueuelen was exceeded: 5038/5000
2012-10-09 20:59:46,970 ERROR zen.zentrap: Discarding oldest 26 events because maxqueuelen was exceeded: 5026/5000
2012-10-09 20:59:56,972 ERROR zen.zentrap: Discarding oldest 22 events because maxqueuelen was exceeded: 5022/5000
2012-10-09 21:00:06,972 ERROR zen.zentrap: Discarding oldest 24 events because maxqueuelen was exceeded: 5024/5000
2012-10-09 21:00:11,973 ERROR zen.zentrap: Discarding oldest 37 events because maxqueuelen was exceeded: 5037/5000
2012-10-09 21:00:16,974 ERROR zen.zentrap: Discarding oldest 35 events because maxqueuelen was exceeded: 5035/5000

zeneventlog:

2012-10-09 20:53:34,671 INFO zen.zeneventlog: 0 devices processed (0 datapoints)
2012-10-09 20:53:34,672 INFO zen.collector.scheduler: Tasks: 1 Successful_Runs: 3502 Failed_Runs: 0 Missed_Runs: 8 Queued_Tasks: 0 Running_Tasks: 0
2012-10-09 20:58:36,921 INFO zen.zeneventlog: 0 devices processed (0 datapoints)
2012-10-09 20:58:36,952 INFO zen.collector.scheduler: Tasks: 1 Successful_Runs: 3502 Failed_Runs: 0 Missed_Runs: 8 Queued_Tasks: 0 Running_Tasks: 0
2012-10-09 20:58:36,954 INFO zen.zeneventlog: 0 devices processed (0 datapoints)
2012-10-09 20:58:36,954 INFO zen.collector.scheduler: Tasks: 1 Successful_Runs: 3502 Failed_Runs: 0 Missed_Runs: 8 Queued_Tasks: 0 Running_Tasks: 0
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/68968#68968]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Shane Scott
2012-10-10 14:30:47 UTC
Permalink
Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/68981#68981

--------------------------------------------------------------
sanane:

High event queues mean that you potentially don't have enough zenhub workers to clean up invalidation messages. Monitor the hub queue length. If it's double or more than a few  than the worker count you'll need more workers. The zenperfsnmp cycle time being exceeded means that the cycle time is not long enough to poll all the devices. You may need to break your devices up across more than one poller, decrease the items polled, decrease the snmp timeouts and tries (if you have many devices that are sporadic) or increase the polling cycle interval. You can also try increasing the maxOids zProperty, but be aware that some devices do not like high snmp query rates. This is especially true of Cisco ASA and CSS appliances.

Best,
--Shane Scott (Hackman238)
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/68981#68981]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-12 07:53:32 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69047#69047

--------------------------------------------------------------
THX for you reply Shane.

i did the following:

in zenhub.conf

workers 40
cachesize 100000
pcachesize 2000

in zenprocesss.conf
monitor      localhost
## cacheconfigs true ( zenoss has commented this line out when rebooted the zenoss-stack, why ??? )
maxqueuelen 10000
eventflushcunksize 200



back to your story:

* High event queues mean that you potentially don't have enough zenhub workers to clean up invalidation messages.

i firstly had no workers, then i added them and it is now a bit better, i am seeing not so many gaps as before


* Monitor the hub queue length. If it's double or more than the worker count you'll need more workers.

how can i monitor this?


* The zenperfsnmp cycle time being exceeded means that the cycle time is not long enough to poll all the devices. You may need to break your devices up across more than one poller, decrease the items polled, decrease the snmp timeouts and tries (if you have many devices that are sporadic) or increase the polling cycle interval.

how can i break into more than one poller? i already tried to decrease and in crease the value of the cycle time from 100 to 500 seconds...



* You can also try increasing the maxOids zProperty, but be aware that some devices do not like high snmp query rates. This is especially true of Cisco ASA and CSS appliances.
how???
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69047#69047]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
jcurry
2012-10-12 08:12:47 UTC
Permalink
jcurry [http://community.zenoss.org/people/jcurry] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69049#69049

--------------------------------------------------------------
Sanane,
We have just been through something very similar with unexplained gaps in graphs - random on some devices and often around the same time of day.  We were worries about load on the Zenoss server so have significantly increased resources - especially memory.  Our gaps have vanished since doing this.  The messages you are getting suggests your system is resource stressed too.

Cheers,
Jane
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69049#69049]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-12 08:35:35 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69051#69051

--------------------------------------------------------------
Thx Jane,

I got a HP proliant server with 8gb of memmory... I think that must be enough?
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69051#69051]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Ryan Matte
2012-10-12 08:56:54 UTC
Permalink
Ryan Matte [http://community.zenoss.org/people/rmatte] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69071#69071

--------------------------------------------------------------
First off, 40 workers is overkill for zenhub.  Don't use more workers than the number of cpu cores on the box.  Also, if you navigate to Advanced -> Collectors -> Localhost -> Performance and scroll down to the graph that shows the number of datapoints being collected, how many are there?  I currently have a server that is collecting over 70k snmp datapoints and I've thrown 16GB of RAM at it to make sure it's snappy.  Also perform a "free -m" at the command line to make sure you aren't biting in to swap and an "iostat" to check your I/O wait time.  If the wait time percentage has a high percent (above 4 of 5 percent) then your disk I/O is inadequate (or you need to tune your filesystem options to improve it).
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69071#69071]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-14 18:21:13 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69151#69151

--------------------------------------------------------------
@ RyAN :

There are almost 20k data points created. There is 8GB of RAM available.

-what do you mean with the "free -m " .
-when installing ubuntu, i did slected some swap,

@Shane

i am running 3.2.1

i got a 8 Cpu and turned the workers back to 8 in the zenhub.conf

i set the snmp polling interval in Advanced -> Collectors -> Localhost -> EDIT...

i got a ubuntu 11 desktop version. there are just packs installed for Juniper switches and firewall, and fping.
HP monitoring(which i am not using now).


| ZenPacks.BlakeDrager.fping (https://10.66.126.10/zport/dmd/ZenPackManager/packs/ZenPacks.BlakeDrager.fping) | BlakeDrager | Blake Drager | 1.0 | Yes |
| | ZenPacks.ZenSystems.Juniper (https://10.66.126.10/zport/dmd/ZenPackManager/packs/ZenPacks.ZenSystems.Juniper) | ZenSystems | Jane Curry | 1.2 | Yes |
| | ZenPacks.community.HPMon (https://10.66.126.10/zport/dmd/ZenPackManager/packs/ZenPacks.community.HPMon) | community | Egor Puzanov | 2.2.72 | Yes |
| | ZenPacks.community.deviceAdvDetail (https://10.66.126.10/zport/dmd/ZenPackManager/packs/ZenPacks.community.deviceAdvDetail) |

@Shane and Ryan
some of the device has beter graphs some not... how is dat possible?

i have added 2 pictures... two devices... one with just 1 gap in the last 4 hours

Loading Image... Loading Image...



and the other... well see it yourself
Loading Image... Loading Image... :( ((((((
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69151#69151]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Ryan Matte
2012-10-15 13:06:45 UTC
Permalink
Ryan Matte [http://community.zenoss.org/people/rmatte] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69164#69164

--------------------------------------------------------------
Ok, so first off, free -m is a command that you execute to see how much memory you are using...

For instance...

[***@lab02 ~]$ free -m
             total       used       free     shared    buffers     cached
Mem:          3829       *3193*        636          0        164        358
-/+ buffers/cache:       2670       1158
Swap:         1023          *0*       1023

That shows me that I'm using roughly 3.2GB of memory and 0MB of swap.  When it comes to Zenoss you never ever want your server touching swap space, it completely kills performance.

Another thing to check is iostat...

[***@lab02 ~]$ iostat
Linux 2.6.32-279.9.1.el6.x86_64 (lab02.novanoc.com)     15/10/12     _x86_64_    (1 CPU)

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
          1.97    0.00    0.37    *2.61*    0.00   95.05

Device:            tps   Blk_read/s   Blk_wrtn/s   Blk_read   Blk_wrtn
scd0              0.00         0.00         0.00        288          0
sda               6.26         0.59       117.04     647160  128394098
dm-0             14.65         0.58       117.04     638106  128394096
dm-1              0.00         0.00         0.00       2376          0

What you're interested in is the %iowait value.  In this case mine is 2.61 which is decent.  You ideally want it under 5%.

Another possible issue could be your network connection to some of the devices.  If you have a slow/unreliable network connection to them it'll translate in to gaps in graphs as well since Zenoss won't be able to collect all of the data that it needs to within the 5 minute cycle.

8GB of RAM is really the bare minimum for Zenoss 4, and with the number of OIDs that you are collecting it wouldn't surprise me if you're biting in to swap a bit.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69164#69164]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 14:33:30 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69165#69165

--------------------------------------------------------------
|          | total   | used   | free | shared | buffers | cached |
| Mem:      | 8064   | 7657    | 406      | 0    | 188   | 4959 |
| -/+ buffers/cache:   | 2509   | 5554 |
| Swap:     | 8181    | 153   | 8028 |


so it is using swap :(
how can i unroll this?


| Linux 3.2.0-24-generic-pae (server-ProLiant-DL160-G6) | 10/15/2012 | _i686_ | (8 CPU) |

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
|        | 1.87 | 0.00 | 0.30   22.12 | 0.00   75.71 |

| Device:        | tps | kB_read/s | kB_wrtn/s | kB_read | kB_wrtn |
| sda         | 290.22   | 116.16  | 2921.26   57250291 1439758160 |


how is this even possible 22%
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69165#69165]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Ryan Matte
2012-10-15 14:38:12 UTC
Permalink
Ryan Matte [http://community.zenoss.org/people/rmatte] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69179#69179

--------------------------------------------------------------
So you're using swap space (which is going to affect your disk I/O), and even without swap you'd probably be sitting at around 15% meaning your storage is not fast enough to keep up with the required I/O).  Writing to RRD files is very I/O intensive.  Your hardware just isn't going to cut it as is I'm afraid.  You need more RAM (so that you can increase the amount of memory allocated to MySQL and memcache to take some load off your disks) and faster storage.

To clear the swap back in to memory for now you can do the following as root:

sync && sync && swapoff -a && swapon -a && sync && sync
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69179#69179]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 15:00:25 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69167#69167

--------------------------------------------------------------
it dropped from 23 to 20 % .. iowait

the swappiness set with the command sysctl vm.swappiness=10 ( UBUNTU ) and the rebooted
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69167#69167]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
omeganon
2012-10-15 15:12:53 UTC
Permalink
omeganon [http://community.zenoss.org/people/omeganon] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69168#69168

--------------------------------------------------------------
Post by Ryan Matte
So you're using swap space (which is going to affect your disk I/O), and even without swap you'd probably be sitting at around 15% meaning your storage is not fast enough to keep up with the required I/O). 
He's using 153M of swap; that's trivial and could be the result of some long unused process being swapped out. The real picture is that he has almost 5G of memory dedicated to filesystem cache. If the kernel can dedicate that amount of memory to cache, he isn't under memory pressure at all.

Based on the systems stats provided, I see no system related problems at all.

The only real way to determine if the disks are overburdened (and I doubt they are looking at that number), is to run something like 'iostat -x 1' for a period of time (or use Zenoss to graph it =) ) and see what the utilization % is. Be aware that if the disk is a hardware raid of some type that iostat will overstate the utilization.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69168#69168]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 15:18:51 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69169#69169

--------------------------------------------------------------
@Omeganon

after i did al the operations
result is:


|          | total   | used   | free | shared | buffers | cached |
| Mem:      | 8064   | 3178   | 4885      | 0    | 128   | 1343 |
| -/+ buffers/cache:   | 1707   | 6356 |
| Swap:     | 8181      | 0   | 8181 |


| Linux 3.2.0-24-generic-pae (server-ProLiant-DL160-G6) | 10/15/2012 | _i686_ | (8 CPU) |

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
|        | 5.34 | 0.08 | 0.54   19.51 | 0.00   74.53 |

| Device:        | tps | kB_read/s | kB_wrtn/s | kB_read | kB_wrtn |
| sda         | 263.08  | 1023.89  | 2374.75 | 1508179 | 3497984 |
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69169#69169]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
omeganon
2012-10-15 15:26:34 UTC
Permalink
omeganon [http://community.zenoss.org/people/omeganon] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69170#69170

--------------------------------------------------------------
Swap use and vm.swappiness was a misdirection for you. You have no apparent swap use issues and the fact that you may be using swap isn't automatically a problem. There are common cases where you may be using swap that have zero impact on the performance of your system.

Run 'iostat -x 1' for a period of time and watch %util. If it stays near 100%, your disk can't perform at the iops level that you need it to. If you have many data points, that wouldn't be unexpected for a single disk to be insufficient.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69170#69170]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 15:31:59 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69171#69171

--------------------------------------------------------------
| |
avg-cpu:  %user   %nice %system %iowait  %steal   %idle
|        | 0.88 | 0.00 | 0.38   47.92 | 0.00   50.82 |

| Device:     | rrqm/s   wrqm/s | r/s | w/s | rkB/s | wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util |
| sda           | 0.00 | 27.00 | 0.00  261.00 | 0.00  1616.00 | 12.38   145.94  840.92 | 0.00  840.92   3.83 100.00 |


util is at 100%...
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69171#69171]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
omeganon
2012-10-15 15:49:41 UTC
Permalink
omeganon [http://community.zenoss.org/people/omeganon] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69172#69172

--------------------------------------------------------------
So, based on this single snapshot, you don't have sufficient disk IOP's for what you're asking it to do. There are two solutions --

1) ask your disks to do less by reducing what you're collecting or reducing the frequency that you're collecting it at or
2) increase the IOPs available to meet your demand. The only way to do this is to increase the number of disks by splitting the data over multiple disks via mounts (inefficient to do with Zenoss) or by using a multi-disk RAID (i.e. 4-6 disk Raid 10). You could use software RAID but I generally prefer a hardware RAID controller with battery backed write caching to get even more write performance + resiliency.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69172#69172]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
jmp242
2012-10-15 15:51:48 UTC
Permalink
jmp242 [http://community.zenoss.org/people/jmp242] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69173#69173

--------------------------------------------------------------
omeganon is right, RAID is likely the way to go. I do suggest starting with Software RAID unless you already have the expensive sort of RAID card omeganon is talking about. I personally prefer software RAID myself...

--
James Pulver
ZCA Member
LEPP Computer Group
Cornell University
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69173#69173]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Shane Scott
2012-10-15 15:57:31 UTC
Permalink
Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69175#69175

--------------------------------------------------------------
The problem is less raw IO and more IO overhead and queue depth clearance. Writting a single file of 10 gigabytes vs 10GB of 60 kilobytes of files that add up to 10 gigabytes arequire vastly differnt amounts of IO attention. Optimizing mount parameters and using the smallest cluster size can help, as I've recently shown rmatte on ext4, but it won't solve the problem. You'll need a RAID, ideally. I reccomend a RAID 10 of commidity SSD disks. Use XFS with a cluster size of 512  bytes or ext4 with the minimum cluster size and commit=100. If your RAID has battery backed cache, also use barrier=0. There are more options as well, but those are the cost effective ones. At Rackspace we use Fusion IO cards in RAID to get the performance we need on very large rollouts (20K devices)

Best,
--Shane Scott (Hackman238)
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69175#69175]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 18:35:09 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69207#69207

--------------------------------------------------------------
Loading Image... Loading Image...
what could i change to this?

when i change the snmp interval cycle i get the error:
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69207#69207]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Ryan Matte
2012-10-15 15:58:23 UTC
Permalink
Ryan Matte [http://community.zenoss.org/people/rmatte] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69191#69191

--------------------------------------------------------------
Also, since he's already using all of his memory he has no additional memory to allocate to memcache or to MySQL, so his system is performing more reads from disk than it needs to.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69191#69191]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
omeganon
2012-10-15 16:14:29 UTC
Permalink
omeganon [http://community.zenoss.org/people/omeganon] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69193#69193

--------------------------------------------------------------
Post by Ryan Matte
Also, since he's already using all of his memory he has no additional memory to allocate to memcache or to MySQL, so his system is performing more reads from disk than it needs to.
But he _wasn't_ using all of his memory. He had over 60% of it used as filesystem cache, which the kernel will automatically recover/use on an as-needed basis. He has plenty of free memory. Additionally, from the numbers he provided, 90% of his disk activity is writes, as expected, and no amount of allocation to memcached or MySQL is going to help that.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69193#69193]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Ryan Matte
2012-10-15 16:19:56 UTC
Permalink
Ryan Matte [http://community.zenoss.org/people/rmatte] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69194#69194

--------------------------------------------------------------
Yeah, you're right, I missed that.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69194#69194]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 16:30:39 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69195#69195

--------------------------------------------------------------
so guys what do you recommonded.... buy a extra HDD?

i already have a HP 2TB SATA HDD, 3G, 7200rpm, 3.5", LFF...
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69195#69195]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Ryan Matte
2012-10-15 16:31:59 UTC
Permalink
Ryan Matte [http://community.zenoss.org/people/rmatte] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69184#69184

--------------------------------------------------------------
It's not the size that's the issue, it's the write speed (the RPMs basically).  By adding a second HD you double your write speed.  That's why using RAID 5 or RAID 10 is so efficient.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69184#69184]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 16:43:03 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69196#69196

--------------------------------------------------------------
i have restarted zenoss:

| Device:     | rrqm/s   wrqm/s | r/s | w/s | rkB/s | wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util |
| sda           | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00   0.00   0.00 |

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
|        | 0.37 | 0.00 | 0.37 | 0.00 | 0.00   99.25 |

| Device:     | rrqm/s   wrqm/s | r/s | w/s | rkB/s | wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util |
| sda           | 0.00 | 0.00 | 0.00   38.00 | 0.00   208.00 | 10.95 | 3.01   79.16 | 0.00   79.16   3.37  12.80 |

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
|        | 0.62 | 0.00 | 0.25 | 2.25 | 0.00   96.88 |

| Device:     | rrqm/s   wrqm/s | r/s | w/s | rkB/s | wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util |
| sda           | 0.00 | 16.00 | 0.00 | 4.00 | 0.00 | 80.00 | 40.00 | 0.27   67.00 | 0.00   67.00  25.00  10.00 |

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
|        | 0.38 | 0.00 | 0.25 | 0.00 | 0.00   99.37 |

| Device:     | rrqm/s   wrqm/s | r/s | w/s | rkB/s | wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util |
| sda           | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00   0.00   0.00 |



why am i getting now 0-15 % util ????
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69196#69196]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Ryan Matte
2012-10-15 16:44:22 UTC
Permalink
Ryan Matte [http://community.zenoss.org/people/rmatte] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69197#69197

--------------------------------------------------------------
Because Zenoss hasn't finished fully loading yet most likely.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69197#69197]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 18:37:07 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69202#69202

--------------------------------------------------------------
@Ryan

i am watching it more than 2 hours now... i dont see any diffrence in the disk utilization....i alternates between 0-15%

So i think ... maybe.... the swappines and the Swap did make a diffrence for now i think... because i didnt restart zenoss after the commands.....but i rebooted the system....
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69202#69202]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
omeganon
2012-10-15 18:41:35 UTC
Permalink
omeganon [http://community.zenoss.org/people/omeganon] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69203#69203

--------------------------------------------------------------
Or perhaps you had some other process outside of Zenoss that was doing the bulk of the writing to the disk that didn't restart after reboot. Since the disk is /dev/sda, I assume that this is also your OS disk so it could have been anything...
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69203#69203]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 18:47:40 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69204#69204

--------------------------------------------------------------
no it is Zenoss that is using 100% almost...otherwise it is around 0%
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69204#69204]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 22:18:32 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69223#69223

--------------------------------------------------------------
i left it running for almost 3 hours now...

as Ryan predicted... the iowait is now alternating and jumping around between 8-18 %

but the sda has an util of 100%
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69223#69223]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Shane Scott
2012-10-16 13:37:54 UTC
Permalink
Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69243#69243

--------------------------------------------------------------
sanane:


How's it looking today? do you have a budget for a small RAID with SSD's?

Best,
--Shane Scott (Hackman238)
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69243#69243]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-16 15:47:53 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69259#69259

--------------------------------------------------------------
same... still util 100% en iowait hanging between 10-20 %....

is it possible to filter in zennoss the SNMP....?



i'm going to try set the cycle for 600sec... maybe if that helps....



Another thing.... i got just 30 devices... and it is so annoying that when i want to load the infrastructure is take almost a minute to show the devices....
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69259#69259]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Shane Scott
2012-10-16 15:49:54 UTC
Permalink
Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69249#69249

--------------------------------------------------------------
sanane:

What's your hardware and is the box a VM? I appologize if that's been answered already.

Best,
--Shane Scott (Hackman238)
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69249#69249]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-16 16:19:21 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69260#69260

--------------------------------------------------------------
no not VM just a ubuntu(desktop version) installation on the server...

i got a : HP ProLiant DL160 G6 E5620, 8GB with HP 2TB SATA HDD, 3G, 7200rpm, 3.5", LFF.

ubuntu 12.04
zenoss 3.2.1 (with fping and juniper mib)
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69260#69260]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Shane Scott
2012-10-16 16:27:06 UTC
Permalink
Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69250#69250

--------------------------------------------------------------
sanane:

Just making sure- you're running ubuntu 64-bit, right?

Can you post your my.cnf for your zenoss mysqld server? What mount options and fs are you using for $ZENHOME/perf? Is the mysqld data structure on the same fs as $ZENHOME/perf?

--Shane
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69250#69250]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-16 16:40:09 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69262#69262

--------------------------------------------------------------
ubuntu 32-bit


/usr/local/zenoss/zenoss/etc/my.cnf

# You can set .._buffer_pool_size up to 50 - 80 %
# of RAM but beware of setting memory usage too high
innodb_buffer_pool_size = 256M
innodb_additional_mem_pool_size = 20M


/usr/local/zenoss/mysql/my.cnf is empty

i installed the zenoss on fresh install of ubuntu.. didnt install anything else then just zenoss
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69262#69262]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-16 16:45:52 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69263#69263

--------------------------------------------------------------
i did set the snmp interval cycle in zenoss in the localhost collector to 600 sec and the restarted zenoss.... iowait is nearly 0% and util is aroound 0-10 % but i see it get also sometimes to 80-100%...

what i dont understand is why the localhost gets the event
Error " threshold of zenperfsnp cycle time exceeded: current value 152"
Critical " threshold of zenperfsnp cycle time exceeded: current value 265"
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69263#69263]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Shane Scott
2012-10-16 16:47:22 UTC
Permalink
Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69267#69267

--------------------------------------------------------------
sanane:

Okay, you're going to need to use ubuntu 64bit. Memory above ~3.4GB or so uses PAE and will be really slow. When you partiion the disk, create a /boot, / (root), and a /opt/zenoss/perf. The /boot and / (root) should be ext3, they can be ext4 if the distro supports it. The /opt/zenoss/perf partiion should be ext4. Once installation is finished update the platform by apt. After that install zenoss 3.2.1 stack 64 and start it then stop it. This will do the initial config. Modifying the my.cnf refrenced by the stack installer, set the buffer pool size to be 256 to 512MB. Set the thread concurrency to 2 or 4. We can increase it later if needed and tweak more advanced items then. Edit your fstab and update the mount options for the perf partition to include noatime,nodiratime,data=writeback,nobh
,commit=100,nouser_xattr. Execute tune2fs -o journal_data_writeback /dev/[block device representing partition] then restart the system.

--Shane Scott (Hackman238)
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69267#69267]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-16 16:55:05 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69265#69265

--------------------------------------------------------------
Okay, you're going to need to use ubuntu 64bit. Memory above ~3.4GB or so uses PAE and will be really slow. When you partiion the disk, create a /boot, / (root), and a /opt/zenoss/perf. The /boot and / (root) should be ext3, they can be ext4 if the distro supports it. The /opt/zenoss/perf partiion should be ext4.

*+Why do i have to create a /opt/zenoss/perf ?? do i need to install zenoss here? and how can i do that... i thought zenoss will always be installed in /usr....+*


Once installation is finished update the platform by apt. After that install zenoss 3.2.1 stack 64 and start it then stop it. This will do the initial config. Modifying the my.cnf refrenced by the stack installer, set the buffer pool size to be 256 to 512MB. Set the thread concurrency to 2 or 4.

+*How do i set the thread concurrency in MysQL ?*+

We can increase it later if needed and tweak more advanced items then. Edit your fstab and update the mount options for the perf partition to include noatime,nodiratime,data=writeback,nobh,commit=100,nouser_xattr. Execute tune2fs -o journal_data_writeback /dev/[block device representing partition] then restart the system.

*+This is completely new for me... :D i dont follow you what we are doing here :D+*
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69265#69265]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Shane Scott
2012-10-16 17:16:02 UTC
Permalink
Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69270#69270

--------------------------------------------------------------
sanane:

To save you frustration I think it would be best if you hired a linux sysadmin familliar with RHEL/CentOS to build your instance initally or start by using the virtual appliance. Keep in mind the virtual appliance is for smaller monitoring use cases.

Best,
--Shane Scott (Hackman238)
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69270#69270]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-16 17:13:08 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69281#69281

--------------------------------------------------------------
after Set to 600 sec now(snmp cycle), util is rond 0-10 % and iowait around 0-2 % but i dont see any graphs now... do i have to wait more time ?
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69281#69281]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Ryan Matte
2012-10-15 15:57:13 UTC
Permalink
Ryan Matte [http://community.zenoss.org/people/rmatte] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69174#69174

--------------------------------------------------------------
omeganon, if his disks aren't overburdened at 22% disk wait time then I'd be amazed.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69174#69174]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Shane Scott
2012-10-12 15:50:30 UTC
Permalink
Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69116#69116

--------------------------------------------------------------
sanane:

What version are you running? For your hub workers, start with 4 and we'll work from there. 40 is far to many as others have said- my largest instance with 20,000 devices has 24 workers. For breaking up your collectors, we'll need to know what type of installation you're running and what packs you're using. Where did you set the polling interval?

Best,
--Shane Scott (Hackman238)
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69116#69116]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Ryan Matte
2012-10-15 14:39:04 UTC
Permalink
Ryan Matte [http://community.zenoss.org/people/rmatte] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69180#69180

--------------------------------------------------------------
I also recommend decreasing your swappiness value (google for steps to do this for your Linux distro).
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69180#69180]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
sanane enenand
2012-10-15 14:49:48 UTC
Permalink
sanane enenand [http://community.zenoss.org/people/ebeng] created the discussion

"Re: Gap in graphs"

To view the discussion, visit: http://community.zenoss.org/message/69166#69166

--------------------------------------------------------------
but i dont understand it...

i orded specially one harrdisk of with very the biggest cache and read/write time....
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/69166#69166]

Start a new discussion in zenoss-users by email
[discussions-community-forums-zenoss--***@community.zenoss.org] -or- at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]
Loading...