zmgengraphs makes my debian etch box freeze.

[jun]

Hi there,

My debian etch box has started crashing repeatedly since I upgraded to etch. I have traced the problem to zmgengraphs ran by the crontab of the zimbra user.

When I run it as root there is no issue but when ran with su - zimbra it crashes almost each time.

I edited zmgengraphs to specify $debug=1, and I saw that most of the time (unfortunately not always) the box froze at this point (it freezes 9 times out of 10 with su -zimbra, and 5 times out of 10 at this specific point) :

.
.
.
Generating month rrd db for mta on ALL
DEBUG: /opt/zimbra/logger/db/work/mta.ALL.month.Message_Count.rrd created
Generating month graph for mta on ALL
Generating month rrd db for mta on ALL
DEBUG: /opt/zimbra/logger/db/work/mta.ALL.month.Message_Bytes.rrd created
Generating month graph for mta on ALL
DEBUG: main::sqlExec:  select * from mta_aggregate where period='day' order by period_end desc limit 365
Generating year rrd db for mta on ALL
DEBUG: /opt/zimbra/logger/db/work/mta.ALL.year.Message_Count.rrd created
Generating year graph for mta on ALL
Generating year rrd db for mta on ALL
DEBUG: /opt/zimbra/logger/db/work/mta.ALL.year.Message_Bytes.rrd created
Generating year graph for mta on ALL
DEBUG: main::sqlExec:  select distinct(host) from amavis_aggregate

top at the moment of the freeze gives :

top - 22:49:25 up 36 min,  2 users,  load average: 1.66, 1.40, 0.85
Tasks: 109 total,   3 running, 106 sleeping,   0 stopped,   0 zombie
Cpu(s): 40.5%us, 59.5%sy,  0.0%ni,  0.0%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   1018572k total,   802900k used,   215672k free,    75212k buffers
Swap:  1044216k total,        0k used,  1044216k free,   212864k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
13113 zimbra    25   0  7984 5340 2084 R 27.3  0.5   0:02.45 zmgengraphs
 3837 zimbra    17   0  6540 4424  972 S  1.0  0.4   0:01.76 zmmtaconfig
 3962 zimbra    18   0  417m  45m 3936 S  0.7  4.6   0:01.62 mysqld
 2275 mysql     15   0  124m  18m 5124 S  0.3  1.8   0:01.39 mysqld
13021 zimbra    15   0  2232 1144  852 R  0.3  0.1   0:00.73 top
    1 root      15   0  1944  644  544 S  0.0  0.1   0:00.91 init
    2 root      34  19     0    0    0 S  0.0  0.0   0:00.00 ksoftirqd/0
    3 root      10  -5     0    0    0 S  0.0  0.0   0:00.00 events/0
    4 root      10  -5     0    0    0 S  0.0  0.0   0:00.01 khelper
    5 root      17  -5     0    0    0 S  0.0  0.0   0:00.00 kthread

You can see I'm not quite following the requirements since other services are running on the box, namely drupal with its own mysql server, so I use the LD_LIBRARY_PATH workaround. http://bugzilla.zimbra.com/show_bug.cgi?id=9394

I'm lost here !!! I hope someone can help me sort this out.

Thanks in advance !

[dijichi2]

right for a start you must have some dodgy kernel, or a hardware issue - linux 'freezes/crashes' are almost unheard of otherwise. what happens if you run this without the LD_LIBRARY_PATH hack, and what is exact way you are using it? Setting LD_LIBRARY_PATH can have unforeseen consequences, it is usually used as a last resort.

[jun]

Yes you're right it is most probably a kernel or harware issue, since other people with the same hosting company have these issues also with other software than zimbra.

Unfortunately for me, no one (clients or company) seems to have found out what causes these freezes.

So if anyone here has a genius intuition about what is at play here... I'll rephrase that : if anyone has anything to say about what might cause this kind of freezes with this kind of script... Please share, you'll help a bunch of embarassed people :-)

Also I had crashes before trying to use the LD_LIBRARY_PATH workaround so it's probably not it.

Some of the people who have the same issue as me at this hosting company, as well as some people from tech support there say it might be related to the newer version of glibc. Does anyone has experienced such a thing ?

Well I guess this is not really directly related to zimbra so I'll just shut up now!

Thanks for reading this !

Jun.