segfaults in libProxy on 7.7.0.1.0 ?

[smveloso]

Hi,

  Users have been complaining about "pauses" and "sporadic errors" accessing the internet.

  The dashboard shows nothing relevant on the "Alerts" tab, but I noticed that there are some moments (about 5 times a day) when the number of connected clients drops abruptly, sometimes to zero, and then recovers. The data is collected via snmp using the "MCAFEE-MWG-MIB.stClientCount.0" oid.

  Looking at the error logs (in /opt/mwg/log/mwg-errors) I found entries like those below; the timestamps match those of the drops in the number of connected clients.

mwg-core.errors.log:[2016-11-25 10:43:45.726 +00:00] [Core] [TermSignalReceived] 'McAfee Web Gateway Core version: 7.7.0.1.0 - build: 22207' - child process exited (termsignal='11').

mwg-core.errors.log:[2016-11-25 11:05:24.644 +00:00] [Core] [TermSignalReceived] 'McAfee Web Gateway Core version: 7.7.0.1.0 - build: 22207' - child process exited (termsignal='11').

mwg-core.errors.log:[2016-11-25 14:56:10.013 +00:00] [Core] [TermSignalReceived] 'McAfee Web Gateway Core version: 7.7.0.1.0 - build: 22207' - child process exited (termsignal='11').

mwg-core.errors.log:[2016-11-25 16:35:20.657 +00:00] [Core] [TermSignalReceived] 'McAfee Web Gateway Core version: 7.7.0.1.0 - build: 22207' - child process exited (termsignal='11').

mwg-core.errors.log:[2016-11-25 18:28:02.600 +00:00] [Core] [TermSignalReceived] 'McAfee Web Gateway Core version: 7.7.0.1.0 - build: 22207' - child process exited (termsignal='11').

mwg-core.errors.log:[2016-11-25 18:30:05.978 +00:00] [Core] [TermSignalReceived] 'McAfee Web Gateway Core version: 7.7.0.1.0 - build: 22207' - child process exited (termsignal='11').

  Signal 11 is SIGSEGV.

  dmesg recorded more than a hundred occurrences like this:

[3034797.008397] traps: mwg-core[74787] general protection ip:7f15607e863f sp:7f15203cb280 error:0 in libProxy.so[7f1560501000+4ad000]

  Has anyone seen theses error messages ?

  Thanks and regards,

  Sergio

PS.

I can confirm that the errors began after the upgrade to 7.7.0.1. After upgrading on a Friday, there were no errors during the weekend (very light load) and, on Monday the errors began to show.

I was thinking of a hardware problem (bad memory) but such coincidence makes me suspicious ...

Message was edited by: Sergio Veloso

Added confirmation that the errors began only after the upgrade to 7.7.0.1.

[smveloso]

Another thing I noticed (from "/opt/mwg/log/debug/binary_invoke.log"):

[2016-11-29 10:07:50.155 +00:00] a parent instance of Core forked a new child...

[2016-11-29 10:07:50.155 +00:00]     parent pid : 17703

[2016-11-29 10:07:50.155 +00:00]     child pid  : 185980

[2016-11-29 10:07:50.155 +00:00]     reason     : guardian for running instance

[2016-11-29 10:07:50.155 +00:00] a child instance of Core was forked...

[2016-11-29 10:07:50.155 +00:00]     parent pid : 17703

[2016-11-29 10:07:50.155 +00:00]     child pid  : 185980

[2016-11-29 10:07:50.155 +00:00]     reason     : guardian for running instance

This is logged at the same time of the failure.

[wilmer_ojeda]

Buenas tardes Sergio.

No estoy muy seguro si mi sugerencia te sirva, pero valida el modelo y configuración del dispositivo, ya que las nuevas actualizaciones tienen un alto consumo de memoria RAM y el fabricante recomienda si tu dispositivo tiene 4 Gb llevarlo a 8 Gb. Esto fue lo que logre investigar luego de que hicimos la actualización a la versión 7.6.2.

[smveloso]

Buenas !

Thanks for the reply ! I think though that memory is not the problem, because the appliance has over 100GB.

However, the latest release (7.7.0.3) mentions this among the "known issues":

"""

The core process failed on Web Gateway with term signal 11 due to a problem with the PDF opener,

which was caused by a missing root entry in a trailer section. (1168496)

"""

I think that is the issue I am experiencing.

I'll update as soon as possible and come back to report the results.