Message boards :
Number crunching :
Corruptedlibc.so.6???
Message board moderation
Author | Message |
---|---|
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
Last night one of my Linux crunchers decided to stop crunching... Today, trying to start BOINC manually from the command prompt I get the following "rude" message: 07-Aug-2017 10:22:03 Initialization completed 07-Aug-2017 10:22:03 [---] Running CPU benchmarks 07-Aug-2017 10:22:03 [---] Suspending computation - CPU benchmarks in progress *** buffer overflow detected ***: boinc terminated ======= Backtrace: ========= /lib/x86_64-linux-gnu/libc.so.6(+0x777e5)[0x7fe5187527e5] /lib/x86_64-linux-gnu/libc.so.6(__fortify_fail+0x5c)[0x7fe5187f356c] /lib/x86_64-linux-gnu/libc.so.6(+0x116570)[0x7fe5187f1570] /lib/x86_64-linux-gnu/libc.so.6(+0x115ad9)[0x7fe5187f0ad9] /lib/x86_64-linux-gnu/libc.so.6(_IO_default_xsputn+0x80)[0x7fe5187566b0] /lib/x86_64-linux-gnu/libc.so.6(_IO_vfprintf+0x139b)[0x7fe51872950b] /lib/x86_64-linux-gnu/libc.so.6(__vsprintf_chk+0x84)[0x7fe5187f0b64] /lib/x86_64-linux-gnu/libc.so.6(__sprintf_chk+0x7d)[0x7fe5187f0abd] Help!!!! Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
Can't you just download the file again and replace it? Or is it now currently available? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
There's a lot of changes going on with this forced "update" to kernel 4.10. They are basically forcing you to update from kernel 4.8 to 4.10 and there are a ton of updates that go with the update. I've found it's best not to fight it. It means removing the driver from nVidia and installing the Repository driver before letting them update you to kernel 4.10 and accepting all the updates. The drivers from nVidia don't work very well with 4.10 on my machines. If you just let them have their way everything seems to work OK with 4.10 and the 375.66 driver from the Repository. Just the Memory OC doesn't work with coolbits with the Repository driver. I usually do the CUDA work in Ubuntu 14.04.5 anyway, and so far they aren't messing with that. *PITA* |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
Sadly all this happened behind my back - the first thing I knew was that one PC hadn't reported back for a good few hours, but appeared to be running. That is until I got a bit closer in and hooked up a screen and fired up BOINCmanager - which showed "disconnected", and trying a forced disconnect and restart (of the manager) I got the dreaded "connecting to manager" message until that failed. Checking what was actually running showed boinc wasn't running, so I did a force start from the command prompt to see what error messages I would get.... Let's see what happens when I swap the nVidia driver out and in... Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
The biggest thing is to allow them to install all the current updates. I don't know which version you're running, but for me and Ubuntu 16.04, bad things start happening if you don't install all the updates. |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
Mint 18.1 As I say, I was away from the machine when it all kicked off, and it was "sitting idle" for over 12 hours... Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Well, Mint 18.2 will install kernel 4.10 and driver 375.66 which is current. Otherwise, open the package manager, search for libc6, and select to Reinstall all the libc6 items you have installed...along with installing any outstanding updates. Good Luck ;-) |
ML1 Send message Joined: 25 Nov 01 Posts: 20289 Credit: 7,508,002 RAC: 20 |
Last night one of my Linux crunchers decided to stop crunching... Very unlikely to be a corrupt libc file!... More likely is that something got updated(?) whilst Boinc was running. Or that you've found a bug in the Boinc app for trying to output results. Try restarting Boinc to see if that reoccurs? Are you using custom optimized apps?... Do they match the libc version that you are running? Hopefully happy crunchin', Martin [edit] This detail is likely not needed. Most likely just restarting after an update will have things fixed. But just-in-case of interest: From a commandline I get for one of my systems: $ file /lib64/libc.so.6 [/edit] See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
Hmm, if is a bug then that's some bug - it would appear that it "hit" just as both CPN & SETI were doing a server contact, but why then doesn't a restart clear it? First thing I did was to try a restart :-( After the driver shuffle I also tried a restart :-( Now doing a BIG update so can't do anything to test for a while. I think something must have been updated at a critical moment. For SETI I'm running Perti's GPU app, and the versions listed agree with what's installed, but would that cause the trip to be while (trying to) run the benchmark routine? (Interesting - a new version of libc6 has been delivered...) Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
Not sure why, but the data directory appears to have moved from /var/lib to my home directory without any intervention on my behalf. Now to get it back from whence it came.... Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
ML1 Send message Joined: 25 Nov 01 Posts: 20289 Credit: 7,508,002 RAC: 20 |
Not sure why, but the data directory appears to have moved from /var/lib to my home directory without any intervention on my behalf. Take care if you have more than one installation...? (Or more than one way that Boinc gets started?...) Happy crunchin', Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
My thoughts were heading in that direction..... I know where it SHOULD be, and there appears to be an installation there. So where the chuff-chuff could the other one be hiding???? Time to start digging. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
Stranger and Stranger (that well known law firm)... If I start BOINC from the command prompt at anywhere other than /var/lib/boinc-client I get two messages: cc_config.xml not found, using defaults then after a few seconds I get Another instance of BOINC is running (when system monitor reports no instances of BOINC running before hitting the enter key on the command) If I start boinc from /var/lib/boinc-client I get a massive long response, starting with: Starting BOINC client version 7.6.31 for x86_64-pc-linux-gnu then a few lines of libraries followed by Data directory: /var/lib/boinc-client execv: no such file or directory GPU detection failed. error 512 no usable GPUs found found app_info.xml: using anonymous platform App version refers to missing GPU type NVIDIA Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
My thoughts were heading in that direction..... It got there by itself? riight. The question is what's in the BOINC folder in your Home folder? Does it have a projects folder? Anything in the Projects folder? Does it have a client_state.xml file? If it has a client_state.xml you can search for the <hostid>xxxx</hostid> and see if it matches your machine. If the host ID doesn't match your machine, and there's nothing in the Projects folder, I'd say trash the folder from your Home folder. The problem with not finding your GPUs sometimes happens after installing the Repository driver. I once had to install it a few times before my GPUs were found. In the Drivers tab tell it to use the Open source driver so it will uninstall the other driver, after it installs the open source driver, tell it to switch back to the nvidia driver. Then reboot and see if it finds the GPUs. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
then after a few seconds I getThis is correct. The client starts twice. It does so for GPU detection (whether the GPU is correctly detected or not, when it's not that's not because the client started twice). This message shows in stdoutdae.txt on all main platforms. |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
TBar: Right.... hostid matches the one that SETI reported as reporting last night. (8180654). There's a load of tasks sitting in the seti folder, and a quick glance indicates they are the ones seti thinks that pc has. I've tried the swapping the drivers a couple of times, but that would only affect the running of GPU tasks and not stop running CPU task as well. I've just re-installed BOINC (all bits...), no improvement, still not starting and running Let's try another driver installation... Jord: Thanks for that bit of information. I thought I might be onto something, but obviously not. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
So it would appear the BOINC folder in your Home folder is the valid one the machine has been using? So, what about the Folder in /var/lib? Anything in the Projects folder? Does it have a client_state.xml and a valid hostID? The BOINC in your Home folder Will Not Work if there is also a BOINC in /var/lib. See what is in the folder in /var/lib, it's seems the one in your Home folder was the active BOINC. |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
Apart from the usual folders that home folder is empty (and looks very similar to the home folder on another, working linux pc). Home folder is empty - as it always has been on this PC, var/lib has a boinc-client folder that is "well populated". This PC has always run from the /var/lib folder without any issues until last night and I did do a check to make sure it has the correct read/write/execute permissions. I've done a trawl though to compare folders with a known working pc and there appears to be no difference, and no additional stray bits lurking :-( Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Hmmm, I thought you were saying you had a BOINC in your Home folder; ...but the data directory appears to have moved from /var/lib to my home directory without any intervention on my behalf.If that's not the case, then I'm not sure what your problem may be. I trashed my Mint 18.2 Install and replaced it with Xubuntu because I couldn't get BOINC to start. Sounds as though you're having a similar problem. I don't have that problem anymore, Xubuntu starts just fine ;-) |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
Thanks guys for your help & support - I have just installed Mint 18.2, and am "running it in" on stock for a couple of days (well, at lest until the other side of the next outrage). The "new" version of the PC is 8317875 - one day I might get the tuit out and merge the two.... Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.