Bug#885542: Bug #885542: lxcfs: virtualizing the btime field of /proc/stat screws up process start times
Hi, Am Freitag, den 05.10.2018, 09:23 +0200 schrieb Michael Banck: > Please find attached the proposed NMU upload to stable, adopting the fix > that Ubuntu has had in place for a while. I've #910371 in the meantime for a stretch-pu. Attached is a new debiff with correct (I hope) versioning. -- Michael Banck Projektleiter / Senior Berater Tel.: +49 2166 9901-171 Fax: +49 2166 9901-100 Email: michael.ba...@credativ.de credativ GmbH, HRB Mönchengladbach 12080 USt-ID-Nummer: DE204566209 Trompeterallee 108, 41189 Mönchengladbach Geschäftsführung: Dr. Michael Meskes, Jörg Folz, Sascha Heuer Unser Umgang mit personenbezogenen Daten unterliegt folgenden Bestimmungen: https://www.credativ.de/datenschutzdiff -Nru lxcfs-2.0.7/debian/changelog lxcfs-2.0.7/debian/changelog --- lxcfs-2.0.7/debian/changelog 2017-05-20 11:00:52.0 +0200 +++ lxcfs-2.0.7/debian/changelog 2018-10-09 10:10:16.0 +0200 @@ -1,3 +1,12 @@ +lxcfs (2.0.7-1+deb9u1) stretch; urgency=medium + + * Non-maintainer upload. + * debian/patches/revert-the-virtualization-of-btime-field.patch: New patch, +reverts the uptime virtualiziation, fixing process start times, adopted +from upstream commit 72dd97f7 (Closes: #885542). + + -- Michael Banck Tue, 09 Oct 2018 10:10:16 +0200 + lxcfs (2.0.7-1) unstable; urgency=medium * New upstream version 2.0.7 diff -Nru lxcfs-2.0.7/debian/patches/revert-the-virtualization-of-btime-field.patch lxcfs-2.0.7/debian/patches/revert-the-virtualization-of-btime-field.patch --- lxcfs-2.0.7/debian/patches/revert-the-virtualization-of-btime-field.patch 1970-01-01 01:00:00.0 +0100 +++ lxcfs-2.0.7/debian/patches/revert-the-virtualization-of-btime-field.patch 2018-10-04 23:28:08.0 +0200 @@ -0,0 +1,34 @@ +From 72dd97f7ecfa1e118f4b33b3d694694fcf6fd120 Mon Sep 17 00:00:00 2001 +From: Serge Hallyn +Date: Sun, 18 Jun 2017 14:43:22 -0500 +Subject: [PATCH 1/2] (temporarily?) revert the virtualization of btime field + in /proc/stat + +Closes #189 + +This seems to be responsible for corrupting STIME on processlist +inside containers. Hopefully we can find a reasonable way to fix +both, but compared to unvirtualized btime field, bogus STIME field +is the greater evil here. + +Signed-off-by: Serge Hallyn +--- + bindings.c | 5 + + 1 file changed, 1 insertion(+), 4 deletions(-) + +Index: lxcfs-2.0.7/bindings.c +=== +--- lxcfs-2.0.7.orig/bindings.c lxcfs-2.0.7/bindings.c +@@ -3538,10 +3538,7 @@ static int proc_stat_read(char *buf, siz + continue; + if (sscanf(line, "cpu%9[^ ]", cpu_char) != 1) { + /* not a ^cpuN line containing a number N, just print it */ +- if (strncmp(line, "btime", 5) == 0) +-l = snprintf(cache, cache_size, "btime %ld\n", getreaperctime(fc->pid)); +- else +-l = snprintf(cache, cache_size, "%s", line); ++ l = snprintf(cache, cache_size, "%s", line); + if (l < 0) { + perror("Error writing to cache"); + rv = 0; diff -Nru lxcfs-2.0.7/debian/patches/series lxcfs-2.0.7/debian/patches/series --- lxcfs-2.0.7/debian/patches/series 1970-01-01 01:00:00.0 +0100 +++ lxcfs-2.0.7/debian/patches/series 2018-10-04 23:23:13.0 +0200 @@ -0,0 +1 @@ +revert-the-virtualization-of-btime-field.patch
Bug#885542: Bug #885542: lxcfs: virtualizing the btime field of /proc/stat screws up process start times
severity 885542 serious tags 885542 +patch thanks Hi, Am Freitag, den 28.09.2018, 13:43 +0200 schrieb Michael Banck: > On Wed, 27 Dec 2017 21:38:41 +0100 Paul Slootman wrote: > > As I find having accurate start times of processes far more important > > than having an accurate uptime counter for my containers, as do other > > people as well apparently, I would like to see this reverted until a > > better solution for the btime field /uptime is found. > > Same here, this broke the patroni autopkgtest (see #909532 and https://g > ithub.com/zalando/patroni/issues/811 as well as https://github.com/giamp > aolo/psutil/issues/1344) > > As ci.debian.net is running on LXC containers (on stable apparently) > could this be fixed in stable, please? > > Should this be RC as it breaks unrelated software? I've bumped the severity to serious now. Please find attached the proposed NMU upload to stable, adopting the fix that Ubuntu has had in place for a while. Do you prefer to fix this yourselves or should I go ahead? Michael -- Michael Banck Projektleiter / Senior Berater Tel.: +49 2166 9901-171 Fax: +49 2166 9901-100 Email: michael.ba...@credativ.de credativ GmbH, HRB Mönchengladbach 12080 USt-ID-Nummer: DE204566209 Trompeterallee 108, 41189 Mönchengladbach Geschäftsführung: Dr. Michael Meskes, Jörg Folz, Sascha Heuer Unser Umgang mit personenbezogenen Daten unterliegt folgenden Bestimmungen: https://www.credativ.de/datenschutzdiff -Nru lxcfs-2.0.7/debian/changelog lxcfs-2.0.7/debian/changelog --- lxcfs-2.0.7/debian/changelog 2017-05-20 11:00:52.0 +0200 +++ lxcfs-2.0.7/debian/changelog 2018-10-04 16:23:15.0 +0200 @@ -1,3 +1,12 @@ +lxcfs (2.0.7-1.1) stretch; urgency=medium + + * Non-maintainer upload. + * debian/patches/revert-the-virtualization-of-btime-field.patch: New patch, +reverts the uptime virtualiziation, fixing process start times, adopted +from upstream commit 72dd97f7 (Closes: #885542). + + -- Michael Banck Thu, 04 Oct 2018 16:23:15 +0200 + lxcfs (2.0.7-1) unstable; urgency=medium * New upstream version 2.0.7 diff -Nru lxcfs-2.0.7/debian/patches/revert-the-virtualization-of-btime-field.patch lxcfs-2.0.7/debian/patches/revert-the-virtualization-of-btime-field.patch --- lxcfs-2.0.7/debian/patches/revert-the-virtualization-of-btime-field.patch 1970-01-01 01:00:00.0 +0100 +++ lxcfs-2.0.7/debian/patches/revert-the-virtualization-of-btime-field.patch 2018-10-04 16:23:15.0 +0200 @@ -0,0 +1,34 @@ +From 72dd97f7ecfa1e118f4b33b3d694694fcf6fd120 Mon Sep 17 00:00:00 2001 +From: Serge Hallyn +Date: Sun, 18 Jun 2017 14:43:22 -0500 +Subject: [PATCH 1/2] (temporarily?) revert the virtualization of btime field + in /proc/stat + +Closes #189 + +This seems to be responsible for corrupting STIME on processlist +inside containers. Hopefully we can find a reasonable way to fix +both, but compared to unvirtualized btime field, bogus STIME field +is the greater evil here. + +Signed-off-by: Serge Hallyn +--- + bindings.c | 5 + + 1 file changed, 1 insertion(+), 4 deletions(-) + +Index: lxcfs-2.0.7/bindings.c +=== +--- lxcfs-2.0.7.orig/bindings.c lxcfs-2.0.7/bindings.c +@@ -3538,10 +3538,7 @@ static int proc_stat_read(char *buf, siz + continue; + if (sscanf(line, "cpu%9[^ ]", cpu_char) != 1) { + /* not a ^cpuN line containing a number N, just print it */ +- if (strncmp(line, "btime", 5) == 0) +-l = snprintf(cache, cache_size, "btime %ld\n", getreaperctime(fc->pid)); +- else +-l = snprintf(cache, cache_size, "%s", line); ++ l = snprintf(cache, cache_size, "%s", line); + if (l < 0) { + perror("Error writing to cache"); + rv = 0; diff -Nru lxcfs-2.0.7/debian/patches/series lxcfs-2.0.7/debian/patches/series --- lxcfs-2.0.7/debian/patches/series 1970-01-01 01:00:00.0 +0100 +++ lxcfs-2.0.7/debian/patches/series 2018-10-04 16:23:15.0 +0200 @@ -0,0 +1 @@ +revert-the-virtualization-of-btime-field.patch
Bug#885542: Bug #885542: lxcfs: virtualizing the btime field of /proc/stat screws up process start times
Hi, On Wed, 27 Dec 2017 21:38:41 +0100 Paul Slootman wrote: > As I find having accurate start times of processes far more important > than having an accurate uptime counter for my containers, as do other > people as well apparently, I would like to see this reverted until a > better solution for the btime field /uptime is found. Same here, this broke the patroni autopkgtest (see #909532 and https://g ithub.com/zalando/patroni/issues/811 as well as https://github.com/giamp aolo/psutil/issues/1344) As ci.debian.net is running on LXC containers (on stable apparently) could this be fixed in stable, please? Should this be RC as it breaks unrelated software? Michael -- Michael Banck Projektleiter / Senior Berater Tel.: +49 2166 9901-171 Fax: +49 2166 9901-100 Email: michael.ba...@credativ.de credativ GmbH, HRB Mönchengladbach 12080 USt-ID-Nummer: DE204566209 Trompeterallee 108, 41189 Mönchengladbach Geschäftsführung: Dr. Michael Meskes, Jörg Folz, Sascha Heuer Unser Umgang mit personenbezogenen Daten unterliegt folgenden Bestimmungen: https://www.credativ.de/datenschutz