Re: [cfarm-users] gcc67 back online was: gcc67 down ? was: gcc67 instability?
On Tue, 2017-12-26 at 12:41 +0100, Nicolas Gonzalez via cfarm-users wrote: > Hi, > > gcc67 and gcc68 are not reachable for few days, we sent Ryzen > processors > back to AMD by RMA. > > Sorry for the temporary inconvenience. > > ngonzalez Hi, After contacting AMD about our Ryzen freeze issue they immediately sent us 4 new processors: 2 Ryzen 7 1700X and 2 Ryzen 5 1400 :). So gcc67 is back online with a new processor, thanks to Nicolas and Mehdi for their trip to the datacenter. We ran ryzen-test (1) on it and it didn't freeze the machine. Unfortunately gcc68 motherboard appears to be dead, our hoster tetaneutral.net will provide us with a new AM4 motherboard in the coming weeks so gcc68 should be back online soon. Many thanks to AMD! Sincerely, Laurent (1) https://github.com/suaefar/ryzen-test ___ cfarm-users mailing list cfarm-users@lists.tetaneutral.net https://lists.tetaneutral.net/listinfo/cfarm-users
Re: [cfarm-users] Please cleanup your /home on farm machines
On Thu, Jan 18, 2018 at 6:17 AM, Baptiste Jonglez via cfarm-userswrote: > Hi everyone, > > Let's start the new year with a positive action for the compile farm: home > cleanup! > > The farm machines with the most critical disk usage are: > > - gcc110 (99% used out of 1.6 TB) > - gcc20 (98% used out of 826 GB) > - gcc13 (95% used out of 459 GB) > - gcc112 (92% used out of 1.8 TB) > - gcc16 (85% used out of 459 GB) > > Please check each of these machines: if you have unused data in your home > directory, it should be deleted! I can't speak for others, but I would not mind being subject to a script that removes temporary object files that are over N days old. In fact I would welcome it because I am forgetful at times. I don't know how long a "log running" job is for some users. Maybe you can exclude users who are logged in, and clean objects that meet criteria for the rest. Jeff ___ cfarm-users mailing list cfarm-users@lists.tetaneutral.net https://lists.tetaneutral.net/listinfo/cfarm-users
Re: [cfarm-users] Please cleanup your /home on farm machines
Baptiste Jonglez via cfarm-users wrote on 20180118: > Let's start the new year with a positive action for the compile farm: home > cleanup! Good idea ! > The farm machines with the most critical disk usage are: Done, all less than 36 kilobytes ;-) Also cleaned gcc14 and gcc21 and gcc76 where I knew I had obsolete stellarium and open-phd-guiding things stored. -- Hans ___ cfarm-users mailing list cfarm-users@lists.tetaneutral.net https://lists.tetaneutral.net/listinfo/cfarm-users
Re: [cfarm-users] Please cleanup your /home on farm machines
For those of you like myself who don't have any important data stored on the build machines, I used a fairly crude script to finally clean out all of my work. Even on machines that I may have forgotten that I used: #!/bin/sh if [ $# = 0 ] thenCOUNT=20 elseCOUNT=$1 fi USERNAME=cgraff1 while [ "$COUNT" -lt 130 ] do printf "%s\n" "attempting to clean ${USERNAME}@gcc${COUNT}.fsffrance.org" ssh "${USERNAME}@gcc${COUNT}.fsffrance.org" -f 'rm -rf *' COUNT=$((COUNT + 1)) done On 1/18/18, Bart Van Assche via cfarm-userswrote: > On 01/18/18 03:17, Baptiste Jonglez via cfarm-users wrote: >> Please check each of these machines: if you have unused data in your home >> directory, it should be deleted! > > Hello Baptiste, > > How about /tmp? There are several systems in the cluster with thousands > of files in /tmp. How about installing a cron job that e.g. removes all > files and directories from /tmp that are older than one week? This is > the shell command I use on my own systems to remove files from /tmp that > are older than one day: > > find "${1:-/tmp}" -maxdepth 1 ! -mtime 0 -ls -exec rm -rf {} \; > > Bart. > ___ > cfarm-users mailing list > cfarm-users@lists.tetaneutral.net > https://lists.tetaneutral.net/listinfo/cfarm-users > ___ cfarm-users mailing list cfarm-users@lists.tetaneutral.net https://lists.tetaneutral.net/listinfo/cfarm-users
Re: [cfarm-users] Please cleanup your /home on farm machines
Thanks. My /home is always clean because I work, get results, delete. Maybe I keep a script to recreate whetever I was doing. This is known as "The tragedy of the commons" : when something is free for everyone, some people use up the resources as if they were sea water, resulting in there being nothing left, and whatver was free dries up. Mind u, it's nothing new. Here's a Christmas greeting from our Unix sysadmin (Peter Collinson) from 1987 Disc is full, Disc is full, People need to work, Delete those junk files from the disc whereso'ever they lurk, HEY! Disc is full, Disc is full, The Eagle disc is packed, Then we can go on holiday leaving Eagle to be hacked. (Sing loudly to a well known tune) :) M ___ cfarm-users mailing list cfarm-users@lists.tetaneutral.net https://lists.tetaneutral.net/listinfo/cfarm-users
[cfarm-users] Please cleanup your /home on farm machines
Hi everyone, Let's start the new year with a positive action for the compile farm: home cleanup! The farm machines with the most critical disk usage are: - gcc110 (99% used out of 1.6 TB) - gcc20 (98% used out of 826 GB) - gcc13 (95% used out of 459 GB) - gcc112 (92% used out of 1.8 TB) - gcc16 (85% used out of 459 GB) Please check each of these machines: if you have unused data in your home directory, it should be deleted! The disk usage of farm machines in the last year can be seen here: https://cfarm.tetaneutral.net/munin/disk-year.html#Disk%20usage%20in%20percent Thanks for your help in making the compile farm great again, Baptiste PS: for a quick & dirty overview of your files on those machines, you can use the following one-liner: for i in gcc{110,20,13,112,16}.fsffrance.org; do echo $i; ssh $i ls -lh; echo; done signature.asc Description: PGP signature ___ cfarm-users mailing list cfarm-users@lists.tetaneutral.net https://lists.tetaneutral.net/listinfo/cfarm-users