Closed tianluyuan closed 8 months ago
Seems like this occurs intermittently; as of this writing I'm not seeing as many node failures. Maybe something with the broker?
Possibly an issue at these sites. Is there a way to blacklist these nodes? Setting !regexp("GP-ARGO.*", GLIDEIN_Site)
does not seem to have an effect.
124 PrivNet=GP-ARGO-dsu-backfill.23259464147e
222 PrivNet=GP-ARGO-wichita-backfill.5287b250b431
This seems to work !regexp("Wichita", GLIDEIN_Site) && !regexp("Dakota", GLIDEIN_Site)
On a fairly large fraction of nodes client jobs are getting killed prematurely. Full log error output is below.