https://defect.opensolaris.org/bz/show_bug.cgi?id=16734
Summary: opengrok indexer needs to message properly when it
can't fork
Classification: Development
Product: opengrok
Version: unspecified
Platform: ANY/Generic
OS/Version: OpenSolaris
Status: NEW
Severity: normal
Priority: P3
Component: analyzer
AssignedTo: [email protected]
ReportedBy: [email protected]
CC: [email protected]
--- Comment #0 from Dan Price <[email protected]> 2010-08-05 20:47:13 UTC
---
I did a stupid thing and set my opengrok data directory to be in /tmp,
and then attempted to do a very large indexing run.
This in turn filled up my /tmp, which deprived the JVM of
enough swap backing space to be able to fork in order to
execute the ctags program. I believe that the net result
of this was that opengrok was claiming that various files
were not analyzable, but I am not certain the two messages
are related.
It seems to me that if opengrok fails to fork in order to exec
ctags, that should result in a very clear message to the user...
Sadly I lost the log files for this, but it should be easy to
reproduce, at least on Solaris. I'm not sure how other OS's
VM systems handle this case, but on Solaris at least, you need
sufficient swap space reservation free in order to be able to
fork. In other words, a 2GB JVM needs 2GB of swap to be
reservable (in case the process decides to go COW fault all
of those pages). When you run low on swap, you tend to see
large processes start to fail to fork. You can see this with
truss, which will show fork failing with ENOMEM (which I did
see in this case).
--
Configure bugmail: https://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
_______________________________________________
opengrok-dev mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opengrok-dev
