Keep CRAB out of the AFS home - #16
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What happened
A 8300-job production died after ~6 h with every status query failing at once:
Why AFS was involved at all
DSProd keeps everything on EOS — the checkout,
soft/,LAW_HOME,data/, and since #15 even thelocal scratch. But CRAB rewrites its task cache
~/.crab3on every command, atomically via~/.crab3.<pid>, and$HOMEon lxplus is AFS. So each status query wrote to AFS, and the momentthe AFS token lapsed the whole production lost its ability to poll.
The credential renewal was not the gap:
update_kerberos_ticketis byte-identical to FLAF'supdate_kinit(kinit -Rthenaklog) and runs hourly fromcrab_poll_callback. But renewal canonly extend a ticket that is still valid; once the TGT is gone so is the token, and a production
whose polling depends on AFS dies with it.
Fix
The
crabwrapperenv.shinstalls now gives CRAB a home of its own:Every CRAB invocation goes through that wrapper (it is what
PATHresolves, by design since #3), sosubmit, status and kill are all covered. law passes
--proxyto all three, so CRAB never needs~/.globusfrom the real home either — nothing in a production run touches AFS any more.Also: the Kerberos renewal in
crab_poll_callbackwas silent (verbose=0), which is why the logcould not answer "was renewal running?". It now logs.
Testing
Against the live 5000-job CRAB task, with the wrapper's new
HOME:.crab3(124 B) appears in$DSPROD_CRAB_HOME;~/.crab3mtime stays at 13:09:58, the moment of the crash — unchanged by the query,i.e. the write really moved off AFS.
mkdocs build --strictpasses;run_tools/apply_format.shis a no-op on the diff.