Over the last week or so I've noticed an occasional desktop "hang" of 0-20 secs which I had been blaming on Minefield (Firefox 4.0). I expected it to go away with an eventual update so I wasn't all that bugged by it.
I was in the process of using Dolphin to move some files around on my server (new drives
) and Dolphin started crashing occasionally, usually right after launch. But then sometimes it wouldn't.
Additionally, during my disk re-arranging the copy/move process would get so slow that the time estimate would approach 300+ hours. I blamed this on my server and using btrfs for the first time so I rebooted both the server and the desktop and this seemed to help this issue so I don't know if it's related or not.
This morning, I opened Dolphin (as it had crashed last night) and attempted to start some file moving. This is where it starts to get weird: It crashed as soon as I selected the /mnt directory. This behavior was repeated again and again. After three tries with dolphin, I opened a terminal and attempted to "ls" /mnt and the terminal window just became non-responsive - no errors, not frozen, typed text still shows up, just nothing happens and no prompt. I attempted this a few mores times in other terminals and also using sudo - same results.
I launched ksysguard and found all my "ls" attempts listed as "disk sleep" - odd to say the least as my /mnt is just a subdirectory of my /.
So I listed my mounts to check for sure and I did have (as expected) an NFS share mounted from my server as /mnt/new (the new drives I installed). So I did an "umount" and the response was "not mounted" - weird again! Then I tried "ls" again - same result as above.
Now I'm getting ticked! I renamed /mnt to /mntold and it accepted it. Tried "ls /mntold" same result. Then tried "rmdir /mntold/new" and got "...resource busy." I then tried "lsof" and "fuser /mntold/new" and both commands froze just like the others with "disk sleep" reported in ksysguard!
In case anyone is wondering, the server seems normal and no disks are "sleep"ing. My other NFS mounts using the same exact mount lines in fstab and are all behaving normally.
At this point I've narrowed it down to the NFS mount at /mnt/new but I don't no how to fix it! I restart the nfs server and viola' all the frozen terminals un-froze. Now I could "ls" /mntold but got "ls: cannot access /mntold/new: Stale NFS file handle". Unmounting resulted in "resource busy" so I forced it "-f" and this fixed everything. Subsequent re-mounting seems OK.
I have no clue as to why this happened or if it'll happen again. I wanted to share in case someone else ran into this.
I was in the process of using Dolphin to move some files around on my server (new drives
) and Dolphin started crashing occasionally, usually right after launch. But then sometimes it wouldn't. Additionally, during my disk re-arranging the copy/move process would get so slow that the time estimate would approach 300+ hours. I blamed this on my server and using btrfs for the first time so I rebooted both the server and the desktop and this seemed to help this issue so I don't know if it's related or not.
This morning, I opened Dolphin (as it had crashed last night) and attempted to start some file moving. This is where it starts to get weird: It crashed as soon as I selected the /mnt directory. This behavior was repeated again and again. After three tries with dolphin, I opened a terminal and attempted to "ls" /mnt and the terminal window just became non-responsive - no errors, not frozen, typed text still shows up, just nothing happens and no prompt. I attempted this a few mores times in other terminals and also using sudo - same results.
I launched ksysguard and found all my "ls" attempts listed as "disk sleep" - odd to say the least as my /mnt is just a subdirectory of my /.
So I listed my mounts to check for sure and I did have (as expected) an NFS share mounted from my server as /mnt/new (the new drives I installed). So I did an "umount" and the response was "not mounted" - weird again! Then I tried "ls" again - same result as above.
Now I'm getting ticked! I renamed /mnt to /mntold and it accepted it. Tried "ls /mntold" same result. Then tried "rmdir /mntold/new" and got "...resource busy." I then tried "lsof" and "fuser /mntold/new" and both commands froze just like the others with "disk sleep" reported in ksysguard!
In case anyone is wondering, the server seems normal and no disks are "sleep"ing. My other NFS mounts using the same exact mount lines in fstab and are all behaving normally.
At this point I've narrowed it down to the NFS mount at /mnt/new but I don't no how to fix it! I restart the nfs server and viola' all the frozen terminals un-froze. Now I could "ls" /mntold but got "ls: cannot access /mntold/new: Stale NFS file handle". Unmounting resulted in "resource busy" so I forced it "-f" and this fixed everything. Subsequent re-mounting seems OK.
I have no clue as to why this happened or if it'll happen again. I wanted to share in case someone else ran into this.
Comment