Re: Panic and filesystem corruption
Matt Dillon <[EMAIL PROTECTED]> writes: > How fast a link was the cvsup running through? A 640/384 kbps ADSL link. DES -- Dag-Erling Smorgrav - [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Panic and filesystem corruption
:Matt Dillon <[EMAIL PROTECTED]> writes: :> To date nearly all the reported corruption has been to directories :> and not to file contents. Does this hold for you as well? Only :> the directory was corrupted and not any files? : :Umm, the end of the cvsup log was padded with zeroes, but I don't :think that qualifies as file corruption. As far as I could see, no :files were damaged - only one of the gnats subdirectories. : :DES :-- :Dag-Erling Smorgrav - [EMAIL PROTECTED] Good. Oh, I take it back... one of the earlier bug reports did report file corruption, but it may have been due to something else or related to the directory corruption. How fast a link was the cvsup running through? -Matt To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Panic and filesystem corruption
Matt Dillon <[EMAIL PROTECTED]> writes: > To date nearly all the reported corruption has been to directories > and not to file contents. Does this hold for you as well? Only > the directory was corrupted and not any files? Umm, the end of the cvsup log was padded with zeroes, but I don't think that qualifies as file corruption. As far as I could see, no files were damaged - only one of the gnats subdirectories. DES -- Dag-Erling Smorgrav - [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Panic and filesystem corruption
:Matt Dillon <[EMAIL PROTECTED]> writes: :> How old a kernel are you running? : :Maybe five days old. : :DES :-- :Dag-Erling Smorgrav - [EMAIL PROTECTED] That has all of our fixes to date. It fits the M.O. of the problem I've been trying to track down, but I have yet to be able to reproduce the problem consistently. It's good to know that you were cvsup'ing GNATS and it didn't previously exist. That helps a little. We still haven't been able to narrow it down as to whether the bug is related to softupdates or not. If you have time and can run cvsup/rm-rf in a loop, it would be interesting to see if you can reproduce the crash reliably. cvsup's basic operation is to create directories, then for each file it creates the file under a temporary name, writes it out, closes, and then renames it. At the moment I am focusing on the rename operation as being the possible culprit, but no smoking gun has been found yet. To date nearly all the reported corruption has been to directories and not to file contents. Does this hold for you as well? Only the directory was corrupted and not any files? -Matt To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Panic and filesystem corruption
Matt Dillon <[EMAIL PROTECTED]> writes: > How old a kernel are you running? Maybe five days old. DES -- Dag-Erling Smorgrav - [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Panic and filesystem corruption
: :On Wed, Mar 14, 2001 at 04:39:53PM -0800, Matt Dillon wrote: : :> fsck all of your filesystems from single-user to remove the possibility :> of 'old' corruption (as in 'fsck', not 'fsck -p'). : :So if an fsck -f doesn't bomb out, the filesystem should be in an okay :state? : :- alex Right. Or even fsck -n from multiuser mode if all your filesystems are completely idle and synced up. I do that from cron once a week just to be sure that everything is A.O.K. -Matt To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Panic and filesystem corruption
On Wed, Mar 14, 2001 at 04:39:53PM -0800, Matt Dillon wrote: > fsck all of your filesystems from single-user to remove the possibility > of 'old' corruption (as in 'fsck', not 'fsck -p'). So if an fsck -f doesn't bomb out, the filesystem should be in an okay state? - alex To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Panic and filesystem corruption
:I was cvsupping the GNATS database (the *entire* GNATS database, that :is - I didn't already have a copy) when I got this panic: How old a kernel are you running? Kirk and I have been attempting to locate the filesystem bitmap corruption for months. We've fixed a number of bugs but have yet to find a smoking gun. It is possible that it *has* been fixed but people's filesystems still contain some corruption from previous breakage. The corruption appears to be related to heavy directory activity and may or may not be related to softupdates. We just can't tell. We can't reproduce it. We thought we had found someone who could reproduce the problem at will but it turned out to be his RAID controller screwing up (where even newfs followed by an fsck would generate errors). Yahoo's test machines still occassionally show the problem, but so far it's too infrequent to get a good handle on. fsck all of your filesystems from single-user to remove the possibility of 'old' corruption (as in 'fsck', not 'fsck -p'). -Matt To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message