With the release of 0.11.1, I noticed in testing that if we time out, then the CondorVaultStorerWorker doesn't attempt to put an old token file back in place. That is, if we move in a stored token, try to run condor_vault_storer, and it times out, we don't put the stored token back, and so it's effectively "deleted". This isn't a show-stopper, but it should be fixed.
Perhaps we check the context deadline BEFORE running condor_vault_storer, and if it's expired, we put the file back immediately and THEN return the error.
shreyb
commented
10 months ago
With the release of 0.11.1, I noticed in testing that if we time out, then the CondorVaultStorerWorker doesn't attempt to put an old token file back in place. That is, if we move in a stored token, try to run condor_vault_storer, and it times out, we don't put the stored token back, and so it's effectively "deleted". This isn't a show-stopper, but it should be fixed.
Perhaps we check the context deadline BEFORE running condor_vault_storer, and if it's expired, we put the file back immediately and THEN return the error.