Web Based Configuration GUI - very very slow

Which LTM version are you running?

 

 

Do you see any errors in the log files under /var/log/httpd? Does top show httpd running at high CPU or memory percentages? If you restart httpd (bigstart restart httpd) does the latency decrease?

 

 

Aaron

I ran into this problem using v9.3.1 HF2. Turned out that there was a memory leak in the mcpd process which manifested itself as slow GUI responses. The problem was identified as CR91313.

 

 

To diagnose, I ran top and looked at the Size column. This indicates how much memory the each process is consuming. In my case, mcpd was eating something rediculous like 450MB of memory, and increased with each ConfigSync. Look at the memory usage of the processes and see if something is out of whack. You can type "M" to sort by memory usage.

 

 

I am on 9.4.3

 

There are some errors in the httpd log but they are old and not up to date. Retsarting the httpd makes no difference at all.

 

How about restarting Tomcat (bigstart restart tomcat4) and then httpd (bigstart restart httpd)?

I have restarted Tomcat and it made a bit a difference.

 

Looking at TOP there are all sorts of processes poping up mcpd comes up as 30604

 

Tmm is 966M

 

Tomcat4 Java comes up with 3 processes of 6000+ occasionaly but they soon dissapear.

 

Just cllicked network map and just have a loading prompt. Took 144 seconds to load....

 

 

TOP currenlty is

 

PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME CPU COMMAND

 

1584 root RT 0 966M 966M 964M S 1.1 85.9 29:01 0 tmm

 

11174 root 15 0 1100 1100 784 R 0.5 0.0 0:06 0 top

 

1614 root 15 0 4132 4132 3208 S 0.1 0.3 2:23 0 bigd

 

2716 root 15 0 284 284 236 S 0.1 0.0 0:11 0 rrdshim

 

1 root 15 0 552 552 468 S 0.0 0.0 0:03 0 init

 

2 root RT 0 0 0 0 SW 0.0 0.0 0:00 0 migration/0

 

3 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 keventd

 

4 root 15 0 0 0 0 SW 0.0 0.0 0:46 0 kapmd

 

5 root 39 19 0 0 0 SWN 0.0 0.0 0:00 0 ksoftirqd/0

 

8 root 25 0 0 0 0 SW 0.0 0.0 0:00 0 bdflush

 

6 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kswapd

 

7 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kscand

 

9 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kupdated

 

11 root 24 0 0 0 0 SW 0.0 0.0 0:00 0 khubd

 

16 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kjournald

 

168 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kjournald

 

169 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kjournald

 

170 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kjournald

 

171 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kjournald

 

181 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 loop0

 

189 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kjournald

 

1377 root 15 0 1356 1356 852 S 0.0 0.1 2:02 0 syslog-ng

 

1434 root 15 0 1436 1436 1192 S 0.0 0.1 0:00 0 sshd

 

1456 ntp 15 0 2556 2556 2192 S 0.0 0.2 0:02 0 ntpd

 

1502 root 15 0 940 940 580 S 0.0 0.0 0:00 0 crond

 

1525 root 15 0 336 336 276 S 0.0 0.0 0:00 0 runsvdir

 

1526 root 0 -20 5464 5464 4220 S < 0.0 0.4 0:06 0 overdog

 

1527 root 22 0 456 456 404 S 0.0 0.0 0:00 0 agetty

 

1547 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1548 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1549 root 15 0 308 308 272 S 0.0 0.0 0:00 0 runsv

 

1550 root 15 0 308 308 272 S 0.0 0.0 0:00 0 runsv

 

1551 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1552 root 15 0 316 316 276 S 0.0 0.0 0:00 0 runsv

 

1553 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1554 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1555 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1556 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1557 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1558 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1559 root 15 0 312 312 272 S 0.0 0.0 0:00 0 runsv

 

1560 root 15 0 308 308 272 S 0.0 0.0 0:00 0 runsv

 

 

At first tmm memory struck me as too high. But then I checked my LTM, and it's reporting 1.7GB. The performance charts only show TMM at ~350MB. So I'm not sure what to trust.

No ASM

 

Straight BigIP LT1500 with 2GB of RAM V9.4.3 Build 1.4 final

 

I increased the host memory portion as I though it may have been running out of memory (using b db Provision.extraMB 768)

 

It made little difference.

 

Info from a qkview dump

 

MEMORY STATISTICS --

 

| (Host) Total = 1.100GB Used = 403.1MB

 

| (TMM) Total = 956.0MB Used = 23.09MB

 

 

I have now started a call with F5 support team as this is very odd behaviour. My standby switch (obviously now my active one) is fine and has not even had the above memory increase.

 

 

I think TOP shows tmm memory that is allocated but I don't think it reflects what is actually in use. Qkview seems to be a better option and this is what i have sent off to the support team.

jondyke, you're correct about top showing allocated CPU and memory. The perf graphs should be accurate. For a CLI tool, tmstat also shows the actual CPU cycles and memory TMM is using.

 

 

 

CPU: 0% busy 1% idle 99% sleep

 

 

Memory Allocated

 

22,269,504 / 1,807,745,024

 

 

 

 

Aaron

Hi,

 

I would check your DNS settings...

 

If you have "Display Host Names When Possible" enabled (under preferences on the gui) and the names of pool members/nodes are unresolvable the F5 GUI will sit and wait for a DNS timeout (a few minutes). This happens even if it's not actually displaying the names, but is displaying something that could be resolved in DNS.

 

 

I had a similar problem, and found that some sections of the GUI were faster than others (iRules for example). Eventually, I found that my DNS settings were wrong (pointing at valid DNS servers but on an unreachable LAN). Once I corrected this the GUI started responding quickly again.

 

 

Rob