[10.4 stable] Timeout calling dhcpcd #3481
Conversation
Add a timeout when calling dhcpcd in case it hangs; this prevents the whole DPCManager from hanging and ultimatevely making the watchdog fire and reboot the system. Signed-off-by: Christoph Ostarek <[email protected]> (cherry picked from commit e628c59)
before a timeout of 1000 seconds was used which is longer than the watchdog waits (currently 500 seconds) also switch to time.Duration to make time durations clearer Signed-off-by: Christoph Ostarek <[email protected]> (cherry picked from commit 193147f)
for command executions that run in separate worker thread: set timeout to 1000 seconds (as it was before) for direct command executions: set timeout to 400 seconds in order to not trigger the watchdog and therefore a reboot Signed-off-by: Christoph Ostarek <[email protected]> (cherry picked from commit 4240dd3)
rouming
force-pushed
the
10.4-stable-timeout-calling-dhcpcd
branch
from
be38e83 to
0bfcabc
Compare
|
Difference to the previous version:
|
|
Hmm - the jenkins ztests job is failing on this, and for some reason the Eden pipeline does not run on the release/patch branches. Needs to be investigated. |
|
@eriknordmark I do not see any failures. How can I check those? |
|
Jenkins fails because eve can't unlock vault. Why eden has not been started - I have no idea, since I so not see any errors. I asked @yash-zededa , probably he can help. What I did here is to reverted those 3 patches (basically make this PR bogus) in order to get better understanding will something pass or not. |
rouming
force-pushed
the
10.4-stable-timeout-calling-dhcpcd
branch
from
54cb7f4 to
e569873
Compare
|
I manually kicked of a jenkins run for this (now empty) PR. |
|
Tried the image manuall on sc-supermicro-zc2 and get the same failure (attestation completes successfully but vault does not get unlocked). Will try 10.4.2-kvm-amd64 as well. |
|
I've tried this PR build and various 10.4.X builds. |
|
@eriknordmark so we have a xen regression between 10.4.2 and 10.4.3. I will take a look. |
rouming
force-pushed
the
10.4-stable-timeout-calling-dhcpcd
branch
from
e569873 to
0bfcabc
Compare
|
Difference to the previous version:
|
|
We should merge this, but we do need to figure out why 9.4.3-xen-amd64 has issues with the vault. |
These are commits originally submitted by @christoph-zededa to the master branch, which target the issue, which was observed on the 9.7 version. I would like to merge them to the 10.4 stable in order to ask the customer to switch to more recent and stable version.