limetech Posted July 5, 2013 Share Posted July 5, 2013 Download | Release Notes Finally, I think this fixes the "transport endpoint not connected" issue! You guys who can make this happen, please pound on it to verify bug is truly fixed. I had a method by which I could get it to fail and with -16c it's been solid. If it does happen, it should generate a file in the root of the server file system called "/core". Please copy this file to a disk, e.g., "cp /core /mnt/disk1" and then send me an email and we'll figure out a way for you to get it to me (it's around 40MB in size). Turns out the problem was with linux "fuse" component. The "transport endpoint not connected" message was getting generated because the 'shfs' (user share file system) process was getting terminated. The next file system operation triggered the error message because there was no "endpoint" any more to receive the request. So how was 'shfs' process getting terminated? To figure this out I had to recompile with 'debugging' symbols, make it fail, then recover the core dump file. Next used 'gdb' debugger to analyze the core dump and determined that an "assert" was getting triggered in fuse function "unlock_path()". This led to a finding a similar bug report. Investigating the 'fuse development' tree on sourceforge, I notice, hey, there's a new release available! Looking in the ChangeLog we see: 2013-02-04 Miklos Szeredi <[email protected]> * libfuse: fix crash in unlock_path(). Patch by Ratna Manoj This was a tough one. Also fixed the share "free space calculation" bug introduced in -16b. Link to comment
jinjorge Posted July 5, 2013 Share Posted July 5, 2013 awesome stuff Tom. Will install 16c tonight and report back any issues. J Link to comment
garycase Posted July 5, 2013 Share Posted July 5, 2013 Great to see this resolved. Although I haven't personally had the bug, I've followed those who do ... and know it's been "kicking you" a good bit I presume this means v5.0 is getting CLOSE Hopefully this will free up some of your time so you can (finally) add UPS support and e-mail notifications 8) ... perhaps to v5.1 !! [My suggestion to make e-mail notification simple both for you and the users is to just implement it for <name>@gmail.com ... so everything's already done except the username. Anyone can get a free gmail account ... and there wouldn't be any configuration hassles ] Link to comment
doorunrun Posted July 5, 2013 Share Posted July 5, 2013 Like my "good buddies" at work like to say, "Go ahead, take the rest of the week off." Thanks for all the hard work! Link to comment
mr-hexen Posted July 6, 2013 Share Posted July 6, 2013 just couldn't bring yourself to call it rc17??? Tom, honestly you've been knocking out of the park lately, take some needed rest and let the users test it for you! cheers! Link to comment
aaronwt Posted July 6, 2013 Share Posted July 6, 2013 I guess I should have waited before I started clearing two drives on my unRAID3. At least I was able to put rc16c on my unRAID1 and unRAID2. Link to comment
limetech Posted July 6, 2013 Author Share Posted July 6, 2013 Like my "good buddies" at work like to say, "Go ahead, take the rest of the week off." I'll rest after some reports that this bug is fixed Link to comment
limetech Posted July 6, 2013 Author Share Posted July 6, 2013 just couldn't bring yourself to call it rc17??? Tom, honestly you've been knocking out of the park lately, take some needed rest and let the users test it for you! cheers! Thought of -rc16b-2 Link to comment
speeding_ant Posted July 6, 2013 Share Posted July 6, 2013 Testing now - will let you know how it goes. Link to comment
limetech Posted July 6, 2013 Author Share Posted July 6, 2013 Noticed linux kernel 3.10 has been released, dare I upgrade? [Ducking out of the way of objects being thrown at me] I like this quote: "this release has been pretty typical and not particularly prone to problems," Link to comment
madburg Posted July 6, 2013 Share Posted July 6, 2013 Like my "good buddies" at work like to say, "Go ahead, take the rest of the week off." I'll rest after some reports that this bug is fixed I thought netatalk update was next? Link to comment
limetech Posted July 6, 2013 Author Share Posted July 6, 2013 Like my "good buddies" at work like to say, "Go ahead, take the rest of the week off." I'll rest after some reports that this bug is fixed I thought netatalk update was next? There's no rest for the weary. Link to comment
madburg Posted July 6, 2013 Share Posted July 6, 2013 A lot of time was lost, and you are on a roll, so if it keeps going it's appreciated. Link to comment
limetech Posted July 6, 2013 Author Share Posted July 6, 2013 A lot of time was lost, and you are on a roll, so if it keeps going it's appreciated. The version of netatalk in unRaid is 2.2.3. Upgrading to 2.2.4 is easy, but from the change log: Changes in 2.2.4 FIX: Missing UAM links FIX: Lockup in AFP logout on Fedora 17 FIX: Reset signal handlers and alarm timer after successfull PAM authentication. Fixes a problem with AFP disconnects caused by pam_smbpass.so messing with our handlers and timer. FIX: afpd: Fix a possible problem with sendfile on Solaris derived platforms It doesn't look like there's anything there that would concern us. Really what I want to do is upgrade to 3.0.x or even 3.1.x, but that will require significant testing and it would not be wise to delay a 'stable' release further. Link to comment
speeding_ant Posted July 6, 2013 Share Posted July 6, 2013 Like my "good buddies" at work like to say, "Go ahead, take the rest of the week off." I'll rest after some reports that this bug is fixed I thought netatalk update was next? There's no rest for the weary. Too right. Link to comment
madburg Posted July 6, 2013 Share Posted July 6, 2013 A lot of time was lost, and you are on a roll, so if it keeps going it's appreciated. The version of netatalk in unRaid is 2.2.3. Upgrading to 2.2.4 is easy, but from the change log: Changes in 2.2.4 FIX: Missing UAM links FIX: Lockup in AFP logout on Fedora 17 FIX: Reset signal handlers and alarm timer after successfull PAM authentication. Fixes a problem with AFP disconnects caused by pam_smbpass.so messing with our handlers and timer. FIX: afpd: Fix a possible problem with sendfile on Solaris derived platforms It doesn't look like there's anything there that would concern us. Really what I want to do is upgrade to 3.0.x or even 3.1.x, but that will require significant testing and it would not be wise to delay a 'stable' release further. That is one way to look at it, or adding the last 2.x version at the very least since 3.x will be a while for us until we see it, is another. It could end up being a good update for afp for the time being, or not, only one way to find out. Your choice in the end. Link to comment
garycase Posted July 6, 2013 Share Posted July 6, 2013 Installed RC16c and all looks fine. Just finished doing two parity checks with 16b (one with, one without Cache_Dirs) ... and they were fine; so now updated to RC16c, did a few transfers to confirm everything looks okay, and about to kick off another parity check. Link to comment
rick.p Posted July 6, 2013 Share Posted July 6, 2013 Noticed linux kernel 3.10 has been released, dare I upgrade? [Ducking out of the way of objects being thrown at me] I like this quote: "this release has been pretty typical and not particularly prone to problems," file that with - “There is no reason anyone would want a computer in their home.” — Ken Olson, president, chairman and founder of Digital Equipment Corp. (DEC), “We will never make a 32 bit operating system.” — Bill Gates “Flight by machines heavier than air is unpractical (sic) and insignificant, if not utterly impossible.” - Simon Newcomb “The cinema is little more than a fad. It’s canned drama. What audiences really want to see is flesh and blood on the stage.” -– Charlie Chaplin “This ‘telephone’ has too many shortcomings to be seriously considered as a means of communication. The device is inherently of no value to us.” — A memo at Western Union, 1878 “The world potential market for copying machines is 5000 at most.” — IBM, to the eventual founders of Xerox “Fooling around with alternating current is just a waste of time. Nobody will use it, ever.” — Thomas Edison “Television won’t be able to hold on to any market it captures after the first six months. People will soon get tired of staring at a plywood box every night.” — Darryl Zanuck Link to comment
BillyJ Posted July 6, 2013 Share Posted July 6, 2013 Free Space on user shares still seems to behave the same as rc16b. Link to comment
langrock Posted July 6, 2013 Share Posted July 6, 2013 Seems that this indeed did the trick. I can now perform a backup operation to an external drive using rsync, just as it was possible in the good old days of 4.x. Thanks so much for hunting down this annoying bug. I'll test next that the Crashplan module works again as it used to as well. Link to comment
limetech Posted July 6, 2013 Author Share Posted July 6, 2013 Free Space on user shares still seems to behave the same as rc16b. What I fixed was this: in calculating free space for a user share, rc16b only tallied free space on each disk where the share currently exists. In rc16c it also includes free space on each disk where it could potentially exist based on include/exclude masks, and whether it's marked cache-only or split=0. I think there is more work to be done however, in handling the used/free space on the cache disk. Link to comment
limetech Posted July 6, 2013 Author Share Posted July 6, 2013 Seems that this indeed did the trick. I can now perform a backup operation to an external drive using rsync, just as it was possible in the good old days of 4.x. Thanks so much for hunting down this annoying bug. I'll test next that the Crashplan module works again as it used to as well. Yes indeed it was an annoying bug, thank you for the report! Link to comment
BillyJ Posted July 6, 2013 Share Posted July 6, 2013 Free Space on user shares still seems to behave the same as rc16b. What I fixed was this: in calculating free space for a user share, rc16b only tallied free space on each disk where the share currently exists. In rc16c it also includes free space on each disk where it could potentially exist based on include/exclude masks, and whether it's marked cache-only or split=0. I think there is more work to be done however, in handling the used/free space on the cache disk. Share Movies Split Level = 0 Included disks = disk1-disk9 (disk1, disk2, disk3 etc...) Excluded disks = blank Use Cache = Yes Free Space 5.7TB Share Music Split Level = 0 Included disks = disk1-disk9 (disk1, disk2, disk3 etc...) Excluded disks = blank Use Cache = Yes Free Space 2.19TB Is the issue a legacy one from it being setup in SimpleFeatures? Or is my expectation of Free Space incorrect? Thanks Will Link to comment
Recommended Posts