[OpenAFS] forcing coredumpsize in bosserver
Folks, I submitted a patch which would use the fact that bosserver runs as root to override resource limits and always drop a core. The issue it's intended to address is that often people will start bosserver from contaminated environments (where coredumpsize is limited) and then have a crash, and can provide no data... Is anyone deliberately turning off cores with limit? Would a command line switch to bosserver be acceptable to you in lieu of it? (None is provided in this patch yet but it could be if it mattered) http://gerrit.openafs.org/#change,1959 -- Derrick ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] forcing coredumpsize in bosserver
Derrick Brashear wrote: Folks, I submitted a patch which would use the fact that bosserver runs as root to override resource limits and always drop a core. The issue it's intended to address is that often people will start bosserver from contaminated environments (where coredumpsize is limited) and then have a crash, and can provide no data... Is anyone deliberately turning off cores with limit? Would a command line switch to bosserver be acceptable to you in lieu of it? (None is provided in this patch yet but it could be if it mattered) http://gerrit.openafs.org/#change,1959 I'd don't care if core files get generated, but I'd want some way to make sure that the core file doesn't fill the disk partition when it gets written. I'm not sure how large the files can get, but I have had that problem in the past (on other software) with multiple GB core files completely filling a smallish / or /usr partition (as on a file server, I'd want to have as much space as possible go to the vice partitions.) Would it be possible to reserve some space in the vice partition and have the core file written there instead? (I suppose you could argue that could be worse, but in general I have much larger vice paritions than system ones.)
Re: [OpenAFS] forcing coredumpsize in bosserver
On Mon, May 17, 2010 at 11:08 AM, Christopher D. Clausen wrote: > Derrick Brashear wrote: >> >> Folks, >> >> I submitted a patch which would use the fact that bosserver runs as >> root to override resource limits and always drop a core. The issue >> it's intended to address is that often people will start bosserver >> from contaminated environments (where coredumpsize is limited) and >> then have a crash, and can provide no data... >> >> Is anyone deliberately turning off cores with limit? Would a command >> line switch to bosserver be acceptable to you in lieu of it? (None is >> provided in this patch yet but it could be if it mattered) >> >> http://gerrit.openafs.org/#change,1959 > > I'd don't care if core files get generated, but I'd want some way to make > sure that the core file doesn't fill the disk partition when it gets > written. I'm not sure how large the files can get, but I have had that > problem in the past (on other software) with multiple GB core files > completely filling a smallish / or /usr partition (as on a file server, I'd > want to have as much space as possible go to the vice partitions.) > > Would it be possible to reserve some space in the vice partition and have > the core file written there instead? (I suppose you could argue that could > be worse, but in general I have much larger vice paritions than system > ones.) What happens if it's a dbserver and there are no vice partitions? -- Derrick ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] forcing coredumpsize in bosserver
On May 17, 2010, at 16:57 , Derrick Brashear wrote: > Folks, > > I submitted a patch which would use the fact that bosserver runs as > root to override resource limits and always drop a core. The issue > it's intended to address is that often people will start bosserver > from contaminated environments (where coredumpsize is limited) and > then have a crash, and can provide no data... > > Is anyone deliberately turning off cores with limit? yes, we are. Typical servers have more virtual memory than would fit onto any filesystem except /vicepx if fully occupied by the server process. In addition, core dumps tend to be a security hazard, hence 0 is the default limit for any process here. I do see why your preferences are different though ;-) > Would a command > line switch to bosserver be acceptable to you in lieu of it? Fine. - Stephan -- Stephan Wiesand DESY -DV- Platanenenallee 6 15738 Zeuthen, Germany ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] forcing coredumpsize in bosserver
On Mon, May 17, 2010 at 11:40 AM, Stephan Wiesand wrote: > On May 17, 2010, at 16:57 , Derrick Brashear wrote: > >> Folks, >> >> I submitted a patch which would use the fact that bosserver runs as >> root to override resource limits and always drop a core. The issue >> it's intended to address is that often people will start bosserver >> from contaminated environments (where coredumpsize is limited) and >> then have a crash, and can provide no data... >> >> Is anyone deliberately turning off cores with limit? > > yes, we are. Typical servers have more virtual memory than would fit onto any > filesystem except /vicepx if fully occupied by the server process. In > addition, core dumps tend to be a security hazard, hence 0 is the default > limit for any process here. I do see why your preferences are different > though ;-) > >> Would a command >> line switch to bosserver be acceptable to you in lieu of it? > > Fine. The patch as in gerrit has been revised to do so, though I have not tested it yet. -- Derrick ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info