https://issues.apache.org/bugzilla/show_bug.cgi?id=44755
Ruediger Pluem <[EMAIL PROTECTED]> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |NEEDINFO
--- Comment #7 from Ruediger Pluem <[EMAIL PROTECTED]> 2008-04-24 12:39:49 PST
---
(In reply to comment #6)
> Owen, keep in mind that 70 handles are required simply for the output pipe,
> which you already knew. In order to recover gracefully from a stopped logger,
> the 70 read-ends of the pipes are required to pass this to a replacement
We only keep the read ends for the CustomLogs / Transferlogs. We do not do the
same for the error logs. Provided the 70 piped loggers are equally distributed
over CustomLogs / Transferlogs and ErrorLogs this could explain Owens numbers
for 2.2.6:
70 write ends
35 read ends
---
105 fds in total.
The remaining four might be for other purposes (e.g. pipe of death). Of course
this applies only to the main process and not to its childs.
For 2.2.8 with the patches applied this would be
70 write ends
35 read ends
4 don't knows as in 2.2.6
68 additional ones we don't know.
Owen, could you please use lsof which is also available for Solaris and attach
the result filtered for the main process. This may help us a little bit if 68
unknown fd's are read or write end of pipes.
--
Configure bugmail: https://issues.apache.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
