DAOS-16693 telemetry: Avoid race between init/read (#15306) #15322
Conversation
In rare cases, a reader may attempt to access a telemetry node after it has been added to the tree, but before it has been fully initialized. Use an atomic to prevent reads before the initialization has completed. Unlucky readers will get a -DER_AGAIN instead of crashing. Signed-off-by: Michael MacDonald <[email protected]>
|
Ticket title is 'Race between add_metric() and d_tm_get_* causes segfaults' |
mjmac
added
bug
clean-cherry-pick
Yep, just added the label. |
github-actions
bot
added
the
priority
|
Test stage NLT on EL 8.8 completed with status UNSTABLE. https://build.hpdd.intel.com/job/daos-stack/job/daos//view/change-requests/job/PR-15322/2/testReport/ |
mjmac
force-pushed
the
mjmac/DAOS-16693-2.6
branch
2 times, most recently
from
a351d85 to
6e760d1
Compare
Skip-nlt: true Allow-unstable-test: true Required-githooks: true Change-Id: I65b221ee67f200c26e975021d91fa9c6750dfb9b
mjmac
force-pushed
the
mjmac/DAOS-16693-2.6
branch
from
6e760d1 to
b30eb74
Compare
|
Test stage Functional Hardware Medium completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-15322/5/execution/node/1351/log |
|
Skipped NLT due to known problems with it on the branch. There was one hw-medium test failure that appears to be due to hardware issues: This already passed tests on master; I don't see a whole lot of point in grinding away to get it to pass on this branch. Tagging @daos-stack/daos-gatekeeper for forced landing. |
mjmac
added
the
forced-landing
In rare cases, a reader may attempt to access a telemetry
node after it has been added to the tree, but before it
has been fully initialized. Use an atomic to prevent
reads before the initialization has completed. Unlucky
readers will get a -DER_AGAIN instead of crashing.
Signed-off-by: Michael MacDonald [email protected]