Cloudmin interface incredibly slow

Hi, My cloudmin master server had some updates including perl. I ran those updates through yum, but the perl update failed because of an old i386 perl version residing on the system (???). I removed that version and updated the rest (it is a x86_64 server). After those updates (of which i believe perl is/was the most important one concerning cloudmin) the webinterface is slow. And i mean s l o w, especially the left frame. It's working just fine when selecting either virtualmin or webmin. but the cloudmin part of the webinterface cannot be worked with like this. The other yum updates were usermin, wbm-* and wbt-*

How do i fix this, and what could be the problem. A reboot (cough windows solution) did not fix it.

Status: 
Closed (fixed)

Comments

Forgot to mention it's cloudmin version 5.7.

If you run top , which processes are using up the most CPU on your system?

Also, you may need to run /etc/webmin/restart after upgrading perl, as you will have changed the perl binary used by Cloudmin's running webserver..

Top shows nothing special: 14775 root 15 0 184m 99m 2604 S 0.0 0.7 0:00.48 miniserv.pl
7092 root 15 0 117m 97m 56m S 0.0 0.7 0:01.44 clvmd
5605 root 15 0 136m 56m 1928 S 0.0 0.4 0:01.76 miniserv.pl
14352 root 15 0 128m 56m 2952 S 0.0 0.4 0:00.46 status.pl

which while waiting for the result after clicking "cloudmin" link top left.

I did restart webmin, and actually restarted the entire server, without result.

Next things I did so far:

reinstall all perl rpm's (yum reinstall perl-*) remove older versions of webmin still present on the system and reinstall the latest one. Check all the wbm rpm's (yum verify-rpm wbm-*) which gave problems with one. Reinstalled that one.

No result so far. No improvement that is.

For the sake of it, updated the entire server to centOS 5.7. Still the same. Would it be wise / would it help to just reinstall cloudmin? And if so do i keep the settings / virtuals? Because that is about all i can think of now. THere is nothing using up weird resources on the system, there are no weird settings / processes. All seems to work well. But for cloudmin.

Howdy -- reinstalling Cloudmin shouldn't help with the issues that you're seeing.

What is the output of these 3 commands:

uptime
free -m
mailq | tail -1

uptime:

15:37:47 up  4:06,  1 user,  load average: 0.27, 0.25, 0.23

free -m

             total       used       free     shared    buffers     cached
Mem:         13289       3476       9812          0        492        550
-/+ buffers/cache:       2432      10856
Swap:        34175          0      34175

And the mail queue is empty

Would it be possible for one of us to login to your system and see what is going wrong?

Yes that could be arranged. What strikes me as weird though is that if you check free -m, or the mem use of the virtuals (xm list) is that dom0 has about 13G of mem now. That used to be 2G, the amount that was reserved for dom0. That goes for all of the 4 servers in that cluster. All have "some amount" of mem asigned to dom0 (different on all 4 servers), which used to be 2G. The servers itself have 32G mem totally. On top of that there's an error on the cloudmin master server in syslog:

Sep 14 12:13:54 griffoendor kernel: xen_net: Memory squeeze in netback driver.

which, according to google could be fixed by setting a grub boot param:

dom0_mem=2048M

I might give that a try first, but that would imply another reboot which is kind of a hastle on a production server :) I'll get back on it. If that param does not work i'll fix an account for one of you.

dom0 mem back to 2G, but that dind't help. I'm out of clues for now. What kind of access you need? For that i will need your outgoing ip.

I had a look, and it seems like DNS lookup slowness is the cause. Unfortunately Cloudmin 5.7 does a lot of un-necessary lookups, which is fixed in the 5.8 release.

Would you like me to install a 5.8 pre-release on your system?

I had a look, and it seems like DNS lookup slowness is the cause. Unfortunately Cloudmin 5.7 does a lot of un-necessary lookups, which is fixed in the 5.8 release.

Would you like me to install a 5.8 pre-release on your system?

DNS issues is one of the first things i checked as well. For that reason i disabled the server @127.0.0.1 and used our "normal" dns servers. On top of that I have a replication cloudmin server, running the same version (now) and with the same dns settings, and that one is fast like it is supposed to be. But i do think a cloudmin reinstall might help, so installing a pre-release of 5.8 might be a good idea.

Actually, if you can wait a couple of days the 5.8 official release will be out ..

The current version does more DNS lookups than necessary, so any slowness can be magnified significantly.

Updated to 5.8 today, and that seemed to have fixed the slowness. Cloudmin fast and workeable again. Thanx!

Great, I'm glad that helped!

Automatically closed -- issue fixed for 2 weeks with no activity.