Forum Discussion
Terrence
Nimbostratus
NimbostratusManagement Interface has become slow and unresponsive.
As we deploy more and more applications behind our F5 load balancers, the web gui has become slow and unresponsive, sometimes to the point where the gui loses connection to the LTM's. The deployment currently consists of a pair of 3600's in HA running 11.3.0(Build 3131.0), with the following elements deployed:
- 201 virtual servers
- 125 nodes
- 242 pools
- 210 monitors(some might not be active)
- 148 http class profiles
As it stands traffic is not currently being impacted by the issue, however as more objects get moved to the device management becomes more impacted. We are currently peaking at around 10000 active connections, however load does not appear to be the cause as we currently only have ~3000 active connections and the responsiveness is the same.
I know there could be numerous reasons why the management interface is slow, but in the communities experience what is likely to be the cause.
- Stefan_Klotz
Cumulonimbus
Yes I restarted the daemons on the Active device as well. It's a little bit better, before totally unresponsive and now pages are loading correctly but takes some seconds. But on the Standby it loads directly. Is this maybe a known bug and already solved in 11.4.1 HF3, where we could update to? Thank you!
Ciao Stefan :)
Cory_50405Noctilucent
Without diagnosing exactly what the memory leak is that you are experiencing, it's tough to say. I know there are multiple memory leak fixes in 11.3.0 HF8 alone. Those are almost certainly included in v11.4.1 HF3 as well. Either one may help resolve your issue.
I would recommend opening a case with support to diagnose the leak so they could give you a definitive answer.
nitassEmployee
we have the same issue with exactly the same TMOS version 11.3.0. The idea with the memory leakage sound good as our slow response gets slower and slower over time.
memory leak in sod daemon is fixed in 11.3.0 hf8.
ID390090 - Scripts show sod leaks memory
sol14175: BIG-IP 11.3.0 cumulative hotfix
https://support.f5.com/kb/en-us/solutions/public/14000/100/sol14175.htmli am running 11.5.1. the sod daemon has taken about 35M of memory.
e.g.
[root@ve11a:Active:In Sync] config top -cbn 1|grep /usr/bin/sod|grep -v grep 7501 root 20 0 37140 35m 28m S 0.0 0.9 19:08.28 /usr/bin/sod- Stefan_Klotz
Thank you nitass,
we also got the following statement from F5:
The issue is caused by a known bug - "id439637 - Memory leaks in SOD" This happens over time where the failover service SOD increases its memory footprint. There are several cases where the GUI and SSH administration perform poorly as you are experiencing. The SOD daemon will eventually crash or cause other services to crash due to lack of memory resources. The workaround is to reboot the unit once you notice /usr/bin/sod increase in memory size, this temporarily frees the memory and until it happens again.
Ciao Stefan :)
Kris__109062According to http://support.f5.com/kb/en-us/solutions/public/15000/300/sol15330.html SOD memory leak is fixed in 11.3.0 HF9.
I have seen articles that mention it is fixed in 11.3.0 HF8 - http://support.f5.com/kb/en-us/solutions/public/14000/100/sol14175.html
I am running 11.3.0 HF8 and it is still leaking memory which suggests F5 didn't completely fix SOD memory leak in HF8. The affected unit has been up for 174 days and the GUI is already starting to slow down/freeze.
5390 root 20 0 478m 477m 22m S 0.0 14.8 2530:34 /usr/bin/sod
