Issues with CoreDumps location on VCVA

[Nerden]

I am having some issues with the new VCenter Virtual Appliance,

Its currently running in my lab, 4 hosts managed with HA and DRS and so on.

I am getting a problem where my coredumps storage (/storage/core/) is filling up very quickly with 500MB-ish files called core.postgres.# (where # is an incrementally increasing number. I can removed the files in this location and everything continues working fine until this fills up again.

I increased the size of the /storage/core partition using GParted to see what would happen and unsurprisingly it just keeps growing. I don't seem to be able to find any material on the subject using the KB so hopefully someone here can assist.

Thanks in advance

[Nerden]

Re: hostd & core files ?

and

VCSA Coredumps 100%

Seem to be the same problem, but no mention on how to either read/troubleshoot the dumps or stop them

[Borja_Mari]

Hello,

IMHO you have two options:

• Stop and disable the esxi dump collector. Just connect locally to the VCVA and run this commands:

         service vmware-netdumper stop

    chkconfig vmware-netdumper off

• Limit the amount of space used by the cores. In the VCVA, set the maximun size in the NETDUMPER_DIR_MAX variable in the /etc/sysconfig/netdumper file, and restart the service: service vmware-netdumper restart

Best regards,

Pablo

[Nerden]

Fantastic, what does this core.postgres do?

I know PostgreSQL is a database so i assume this is a large amount of data dropped on a regular basis by said database but why?

[Borja_Mari]

Hello,

about the "NETDUMPER_DIR_MAX", take in consideration that the cron process in the VCVA would free space every hour

It seems the vcva's database (postgresql, named by vmware as vPostgre) have some issue.

Perhaps are you over the inventory limit?

VMware KB: vCenter Server in vCenter Server Appliance stops responding when using the embedded datab...

Maybe it will be a good idea to open a case with vmware ...

Best regards,

Pablo

[Nerden]

No where near those limits.

I will take a look in the morning and see what the level is.

I made a note when i changed those settings you advised, 15:44 - 22% (/storage/core/ = 22GB [100GB volume])

Thanks again

[Borja_Mari]

Hello,

you welcome.

Perhaps you are be able to debug the vpostgresql core dumped:

https://wiki.postgresql.org/wiki/Getting_a_stack_trace_of_a_running_PostgreSQL_backend_on_Linux/BSD#...

Best regards,

Pablo

[Nerden]

Pablo, this didn't work, I logged in this morning and low and behold hundreds of core.postgres dumps again taking the disk to 100%

Any more ideas?

[Borja_Mari]

Hello,

IMHO you have some options:

• Open a case with vmware about your issue
• Try to find the reason of this vpostgresql cores.
• As dirty workaround, configure in the VCVA cron that every day remove the oldest vpostgresql cores.

Best regards,

Pablo

[Nerden]

It's an unlicensed lab environment VMWare are likely to tell me to sod off

ill give the other two a go

[Nerden · Accepted solution]

Upgrading to VCVA 5.5.0.1000 (5.5 Update 1) seems to have fixed the issue.

Thanks for all the help Pablo

[Borja_Mari]

Nice to know you have fixed you issue upgrading the VCVA

Best regards,

Pablo