[OpenAFS-Doc] Re: [OpenAFS-devel] vmalloc memory leak in 1.3.84/85?

[Michael D. Norwick]

--On Thursday, July 28, 2005 10:00:04 AM -0700 Miles Davis

you could try enabling pag garbage collection
sysctl -w afs.GCPAGs=1

Huh... I had no idea those were there. The default seems to be 2 -- what
does that mean, and are these documented somewhere?
Documentation on new features? hah.

Gee, seems like I had some dialog on this ohhhh, last week? No
direction has yet been forthcoming.

2 is AFS_GCPAGS_USERDISABLED
only a value of 1 (AFS_GCPAGS_OK) will cause the garbage collector to
run. I think that the point of the non-zero disable value is because
the macro AFS_GCPAGS is used both as a compile time directive (#if
AFS_GCPAGS), but is also the initializer for the run-time variable
afs_gcpags (which is what the sysctl modifies and affects the runtime
behavior of the code when it is compile-time enabled). There are
other possible values for afs_gcpags, which indicate that the code
disabled itself due to encountering an error while examining process
lists.

Michael

[Esther Filderman]

Yep, sorry about that. A minor disaster cropped up [I'd forgotten I
had to give a talk last Saturday!!] and I got swamped with that.

I promise I'll get straight on things this weekend.

e.

Post by Michael D. Norwick

--On Thursday, July 28, 2005 10:00:04 AM -0700 Miles Davis

you could try enabling pag garbage collection
sysctl -w afs.GCPAGs=1

Huh... I had no idea those were there. The default seems to be 2 -- what
does that mean, and are these documented somewhere?
Documentation on new features? hah.

Gee, seems like I had some dialog on this ohhhh, last week? No
direction has yet been forthcoming.

2 is AFS_GCPAGS_USERDISABLED
only a value of 1 (AFS_GCPAGS_OK) will cause the garbage collector to
run. I think that the point of the non-zero disable value is because
the macro AFS_GCPAGS is used both as a compile time directive (#if
AFS_GCPAGS), but is also the initializer for the run-time variable
afs_gcpags (which is what the sysctl modifies and affects the runtime
behavior of the code when it is compile-time enabled). There are
other possible values for afs_gcpags, which indicate that the code
disabled itself due to encountering an error while examining process
lists.

Michael
_______________________________________________
OpenAFS-doc mailing list
https://lists.openafs.org/mailman/listinfo/openafs-doc

[Derrick J Brashear]

Post by Michael D. Norwick

Documentation on new features? hah.

Gee, seems like I had some dialog on this ohhhh, last week? No
direction has yet been forthcoming.

Well, when you figure out how to add hours to the day, I bet you can get
people to tell you about new features. No hours have been forthcoming.