Re: Is this an fsck_ffs memory issue?
On Aug 02 20:55:56, Mark Smith wrote: So yes it is a memory issue, your 500G disk is too big for your Alix1c. To be precise: the 500G _filesystem_ is too big, not the disk, right? fsck'ing 10 separate 50g filesystems would be less of a problem, right? jan
Re: Is this an fsck_ffs memory issue?
Jan Stary wrote: On Aug 02 20:55:56, Mark Smith wrote: So yes it is a memory issue, your 500G disk is too big for your Alix1c. To be precise: the 500G _filesystem_ is too big, not the disk, right? fsck'ing 10 separate 50g filesystems would be less of a problem, right? jan Yes.
Re: Is this an fsck_ffs memory issue?
On Mon, Aug 04, 2008 at 11:00:03AM +0200, M. Feenstra wrote: Thank's for you answer. Yes I have read the FAQ (some people actually do). That is where I came up with the memory as a suspect. The reason I still ask is that is crashes the system. It does not quit with an out of memory kind of error but just freezes the whole device. What I would like to know if this is expected behaviour? You did not answer my previous post. How much swap space have you allocated? -0tto
Re: Is this an fsck_ffs memory issue?
M. Feenstra wrote: Thank's for you answer. Yes I have read the FAQ (some people actually do). That is where I came up with the memory as a suspect. The reason I still ask is that is crashes the system. It does not quit with an out of memory kind of error but just freezes the whole device. What I would like to know if this is expected behaviour? it isn't so much a crash as every running task is waiting for some other running task to release some RAM/swap so it can continue. Computers are very patient. Granted, from a user standpoint, it is almost indistinguishable from a crash, as the machine is not doing what you want, and that won't change IF that is what is going on, at that point, there is really only one big task running, and that task is only going to want more RAM, not give some up. On the other hand, if you have 1G of swap AVAILABLE AT BOOT and 128M RAM, the system should ultimately come back up...but it will take a very long time, as it will be in swap-hell for an hour or more (or less). If your swap is on the same spindle as the big partition you are checking, it will be a lot longer than if it is a different spindle. If your swap is not active yet (i.e., swap-to-file on a partition which you need to fsck, not sure what happens if it is specified in fstab), it doesn't help you here. Nick.
Re: Is this an fsck_ffs memory issue?
On Sat, Aug 02, 2008 at 08:09:41PM +0200, M. Feenstra wrote: Hi, I use a small device (Alix 1C based) running flashboot (OpenBSD 4.3) as my home Bacula backup server. The entire system is running in RAM (flashboot does that for me) except of course for the backup storage which is a 500G USB drive. Today I had an power issue causing the system to go down uncleanly. At boot it would not mount the USB drive anymore reporting I should run fsck. This was ofcourse no surprise and I tried running fsck. But it frooze my device: -- [EMAIL PROTECTED] root]# fsck -v /dev/sd0a start /dev/sd0a wait fsck_ffs /dev/rsd0a ** /dev/rsd0a -- and that is it. Nothing more is happening. If I attach the USB drive to the OpenBSD system I used to build the flashboot I have no troubles fsck'ing the disk. What surprices me is that is hangs the whole system. I can not login, reboot, access the kernel debugger or anything. I can only unplug the power and reboot after issueing this command. Is this caused because the system has no enough memory or am I missing something obvious here? fscking a large filesystem consumes lots of memory. If your setup has too little or no swap this can happen. You are overallocating. Now ideally a system should do something sensible when trying to allocate more memory than available swap, but I guess this is a pretty hairy subject. -Otto
Re: Is this an fsck_ffs memory issue?
M. Feenstra wrote: What surprices me is that is hangs the whole system. I can not login, reboot, access the kernel debugger or anything. I can only unplug the power and reboot after issueing this command. Is this caused because the system has no enough memory or am I missing something obvious here? Mark If you read section 14 of the FAQ : A rough guideline is the system should have at least 1M of available memory for every 1G of disk space to successfully fsck the disk. So yes it is a memory issue, your 500G disk is too big for your Alix1c. Now you know ;)
Re: Is this an fsck_ffs memory issue?
Sorry for hijacking your thread, but... What do you guys think about journalized ffs [1]? Can it help in such situations? Is there any chance to get it into OpenBSD kernel? [1] http://mail-index.netbsd.org/netbsd-announce/2008/07/31/msg37.html 2008/8/2, M. Feenstra [EMAIL PROTECTED]: Hi, I use a small device (Alix 1C based) running flashboot (OpenBSD 4.3) as my home Bacula backup server. The entire system is running in RAM (flashboot does that for me) except of course for the backup storage which is a 500G USB drive. Today I had an power issue causing the system to go down uncleanly. At boot it would not mount the USB drive anymore reporting I should run fsck. This was ofcourse no surprise and I tried running fsck. But it frooze my device: -- [EMAIL PROTECTED] root]# fsck -v /dev/sd0a start /dev/sd0a wait fsck_ffs /dev/rsd0a ** /dev/rsd0a -- and that is it. Nothing more is happening. If I attach the USB drive to the OpenBSD system I used to build the flashboot I have no troubles fsck'ing the disk. What surprices me is that is hangs the whole system. I can not login, reboot, access the kernel debugger or anything. I can only unplug the power and reboot after issueing this command. Is this caused because the system has no enough memory or am I missing something obvious here? Mark
