Description of problem: If I use fakeroot on tar like this: fakeroot tar -xf linux-2.6.37.3.tar.bz2 tar is very slow. No cpu nor disk is used (hardly). Package is being unpacked but I was too unpatient to mesure how long would it take ;) the problem doesn't affect ie bzip2 Version-Release number of selected component (if applicable): fakeroot version 1.14.4 tar (GNU tar) 1.25 How reproducible: always Steps to Reproduce: 1. untar something big with tar + fakeroot Reproducible: Steps to Reproduce:
I forgot to say that there is no problem with tar without fakeroot: time tar -xf linux-2.6.37.3.tar real 0m33.677s user 0m0.293s sys 0m4.257s
CC: (none) => thierry.vignaudAssignee: bugsquad => pterjan
I tried with fakeroot tar -xf linux-2.6.37.3.tar and stopped it after 8 minutes. None of the process (tar/faked) had been using much cpu and it had not made big progress 5067 pterjan 20 0 2500 1056 880 S 1 0.0 0:02.79 tar 5060 pterjan 20 0 2472 1012 252 S 0 0.0 0:00.78 faked Looking at strace it seems "normal", tar is sort of worst case for fakeroot. When running as root, tar will do additionnal things: chown the files to their original user and restore original permissions. So tar keeps doing some stat/chown/chmod and each of them means communicating with faked over tcp... $ time fakeroot tar --no-same-owner --no-same-permissions -xf linux-2.6.37.3.tar real 1m25.951s user 0m0.820s sys 0m4.790s (--no-same-owner --no-same-permissions are the default when not running as root)
yeah I've been stracing tar too and I've same conclusions. Fakeroot uses some tcp communication with faked process? Anyway I tried fakeroot + tar under debian and it's ok. I've also tried mandirva's fakeroot and it's broken too.
It is on Mandriva and Mageia because it is built with --with-ipc=tcp I don't know why...
OK we've a solution then
Keywords: (none) => PATCH
Would be nice to know why oden changed it http://svn.mandriva.com/cgi-bin/viewvc.cgi/packages?view=revision&revision=162573
I've build fakeroot by myself without tcp. I'll test it and let You know if there are any other issuses with it.
for me looks ok
@ Pascal Any news on this bug? Does it still exist?
CC: (none) => marja11
we still have the same version of fakeroot since it was imported, so it didn't get fixed. @ Pascal @ Bruno Bruno is the maintainer of fakeroot now, can this bug be re-assigned to him?
CC: (none) => bruno
(In reply to comment #10) > we still have the same version of fakeroot since it was imported, so it didn't > get fixed. > > > @ Pascal > @ Bruno > > Bruno is the maintainer of fakeroot now, can this bug be re-assigned to him? 3 Months later now, reassigning to Bruno If that is wrong: tell me
Assignee: pterjan => bruno
Uploaded the recompiled as suggested above for Mageia 2
Status: NEW => RESOLVEDResolution: (none) => FIXED