microsoft / omi

Open Management Infrastructure
Other
360 stars 114 forks source link

omi server cannot start: failed to change directory #729

Open kanelo75 opened 1 year ago

kanelo75 commented 1 year ago

I have the following error in the omiserver.log when I start the omi server, and the omi server cannot be started.

failed to change directory to: EventId=10005 Priority=FATAL abnormal termination of engine process detected...shutting down server

I check the code. It should have the directory name after "failed to change directory to:"

JumpingYang001 commented 1 year ago

@kanelo75 can you share the whole omiserver.log? also check these commands: cat /etc/*-release;rpm -qa|grep omi;rpm -qa|grep scx;dpkg -l|grep omi;dpkg -l|grep scx;ps -ef|grep omi;ls -al /opt;ls -al /var/opt/omi/run

cinimodmik commented 1 year ago

omilog.txt Hi @JumpingYang001 , I have attached the logs for your advise

JumpingYang001 commented 1 year ago

@cinimodmik it seems the program cannot load rundir, can you run /opt/omi/bin/omiserver -p? The error message hit here: https://github.com/microsoft/omi/blob/32a26de4b1ee8f6f958f5519904cc13ad50a703d/Unix/engine/engine.c#L56

cinimodmik commented 1 year ago

hi @JumpingYang001, i have only running it via scxadmin -start and systemctl start omid. Do i run /opt/omi/bin/omiserver -p directly?

JumpingYang001 commented 1 year ago

@cinimodmik yes, run /opt/omi/bin/omiserver -p directly with root account.

cinimodmik commented 1 year ago

@JumpingYang001, will try and get back

cinimodmik commented 1 year ago

@JumpingYang001, i have attached the results . omilog2.txt

JumpingYang001 commented 1 year ago

@cinimodmik that means root user can get OMI_GetPath, and omi user cannot get OMI_GetPath, it looks strange, I have no idea now. root user get OMI_GetPath values :https://github.com/microsoft/omi/blob/b133548bc064c5874fdba5f5c6f08d11d78deb39/Unix/server/server.c#L387 https://github.com/microsoft/omi/blob/e4d72481fa2f805148c9c8f4d0183b3e2d7814a8/Unix/base/paths.c#L171 when /opt/omi/bin/omiserver -p

BTW, is your box enabled selinux?

kanelo75 commented 1 year ago

yes, the box is enabled selinux if I am not mistaken. Is there anyway we can configure those -p information in some configuration file and let's omi user to retrieve the information.

JumpingYang001 commented 1 year ago

it has built by default, and some values from /etc/opt/omi/conf/omiserver.conf.

cinimodmik commented 1 year ago

Hi @JumpingYang001,

I have also tried previously to temporally set selinux to permissive but it is still the same failure results.

cinimodmik commented 1 year ago

hi @JumpingYang001, any further advise for this issue

JumpingYang001 commented 1 year ago

@cinimodmik after set selinux t permissive do you reboot the machine? we didn't repro the issue, so I suggest you to create a ticket on SCOM support.

cinimodmik commented 1 year ago

Hi @JumpingYang001 , I have not reboot the machine as selinux will be resumed to enforce if i do that. Noted thanks for your help .

kanelo75 commented 1 year ago

@JumpingYang001 is the SCOM support you mention under Microsoft? @cinimodmik have raised this case to Microsoft before, and there is one Microsoft engineer has come to look at it before we raised the issue here, and he seemed to have no idea too. To move it forward, if we raise to Microsoft again, anything we could do in order to speed up to find the root cause.

JumpingYang001 commented 1 year ago

@kanelo75 yes, SCOM support is under Microsoft. If that's the case, you should ping the engineer to contact OMI team.