| Message ID | C8CF870384EBCF4CAB89D2041A5C80B301135242@SHSMSX101.ccr.corp.intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi, On Wed, 30 Jul 2014, Xue, Chendi wrote: > >From 797dbf4f027200bb22ba89d2f616e254321ec023 Mon Sep 17 00:00:00 2001 > From: Chendi Xue <chendi.xue@intel.com> > Date: Tue, 29 Jul 2014 16:45:45 +0800 > Subject: [PATCH 8/8] Adding do_cmd_okfail when calling ceph-crush-location > > Original init-ceph doesn't use do_cmd_okfail when getting osd_location, > which results in osds wrong mapping( verified by using "ceph osd tree" ) Is this when you are using the -a switch, or when you are starting daemons on the local host? The -a stuff isn't well supported any more (and I've half a mind to kill support for it entirely if the world weren't going toward systemd anyway). Anyway, I'm worried about what the _okfail variant will change the behavior here if the location hook script exits with an error for some reason... sage > > Signed-off-by: Chendi Xue <chendi.xue@intel.com> > --- > src/init-ceph.in | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/src/init-ceph.in b/src/init-ceph.in > index 86eb87f..09c7cbc 100644 > --- a/src/init-ceph.in > +++ b/src/init-ceph.in > @@ -327,7 +327,7 @@ for name in $what; do > if [ "${update_crush:-1}" = "1" -o "${update_crush:-1}" = "true" ]; then > # update location in crush > get_conf osd_location_hook "$BINDIR/ceph-crush-location" "osd crush location hook" > - osd_location=`$osd_location_hook --cluster ceph --id $id --type osd` > + osd_location=$(do_cmd_okfail "$osd_location_hook --cluster ceph --id $id --type osd") > get_conf osd_weight "" "osd crush initial weight" > defaultweight="$(df -P -k $osd_data/. | tail -1 | awk '{ print sprintf("%.2f",$2/1073741824) }')" > get_conf osd_keyring "$osd_data/keyring" "keyring" > -- > 1.9.1 > > > Best Regards, > -Chendi > > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > >
On Wed, 30 Jul 2014, Xue, Chendi wrote: > Hi, Sage > > Yes, this happened when I use the -a switch. > > If I use do_cmd instead of do_cmd_okfail to call the location hook here, > is it being more suitable? I think that'll work, yeah! > I think there are still a bunch of people using "/etc/init.d/ceph -a > start" to start ceph deamon, so why leave this small problem in > init-ceph when "-a" is still supported? I think it's fine. The main issue is that the -a mode only works for clusters deployed the 'old' way (not with ceph-deploy or chef or puppet). Anyway, I'll take a patch with do_cmd (provided you tested it and it works for you). :) sage > > > Best Regards, > -Chendi > > -----Original Message----- > From: ceph-devel-owner@vger.kernel.org [mailto:ceph-devel-owner@vger.kernel.org] On Behalf Of Sage Weil > Sent: Wednesday, July 30, 2014 9:03 AM > To: Xue, Chendi > Cc: ceph-devel@vger.kernel.org > Subject: Re: [PATCH]init-ceph: Adding do_cmd_okfail when calling ceph-crush-location > > Hi, > > On Wed, 30 Jul 2014, Xue, Chendi wrote: > > >From 797dbf4f027200bb22ba89d2f616e254321ec023 Mon Sep 17 00:00:00 > > >2001 > > From: Chendi Xue <chendi.xue@intel.com> > > Date: Tue, 29 Jul 2014 16:45:45 +0800 > > Subject: [PATCH 8/8] Adding do_cmd_okfail when calling > > ceph-crush-location > > > > Original init-ceph doesn't use do_cmd_okfail when getting > > osd_location, which results in osds wrong mapping( verified by using > > "ceph osd tree" ) > > Is this when you are using the -a switch, or when you are starting daemons on the local host? The -a stuff isn't well supported any more (and I've half a mind to kill support for it entirely if the world weren't going toward systemd anyway). > > Anyway, I'm worried about what the _okfail variant will change the behavior here if the location hook script exits with an error for some reason... > > sage > > > > > > Signed-off-by: Chendi Xue <chendi.xue@intel.com> > > --- > > src/init-ceph.in | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/src/init-ceph.in b/src/init-ceph.in index > > 86eb87f..09c7cbc 100644 > > --- a/src/init-ceph.in > > +++ b/src/init-ceph.in > > @@ -327,7 +327,7 @@ for name in $what; do > > if [ "${update_crush:-1}" = "1" -o > > "${update_crush:-1}" = "true" ]; then > > # update location in crush > > get_conf osd_location_hook "$BINDIR/ceph-crush-location" "osd crush location hook" > > - osd_location=`$osd_location_hook > > --cluster ceph --id $id --type osd` > > + osd_location=$(do_cmd_okfail > > +"$osd_location_hook --cluster ceph --id $id --type osd") > > get_conf osd_weight "" "osd crush initial weight" > > defaultweight="$(df -P -k $osd_data/. | tail -1 | awk '{ print sprintf("%.2f",$2/1073741824) }')" > > get_conf osd_keyring "$osd_data/keyring" "keyring" > > -- > > 1.9.1 > > > > > > Best Regards, > > -Chendi > > > > -- > > To unsubscribe from this list: send the line "unsubscribe ceph-devel" > > in the body of a message to majordomo@vger.kernel.org More majordomo > > info at http://vger.kernel.org/majordomo-info.html > > > > > >
diff --git a/src/init-ceph.in b/src/init-ceph.in index 86eb87f..09c7cbc 100644 --- a/src/init-ceph.in +++ b/src/init-ceph.in @@ -327,7 +327,7 @@ for name in $what; do if [ "${update_crush:-1}" = "1" -o "${update_crush:-1}" = "true" ]; then # update location in crush get_conf osd_location_hook "$BINDIR/ceph-crush-location" "osd crush location hook" - osd_location=`$osd_location_hook --cluster ceph --id $id --type osd` + osd_location=$(do_cmd_okfail "$osd_location_hook --cluster ceph --id $id --type osd") get_conf osd_weight "" "osd crush initial weight" defaultweight="$(df -P -k $osd_data/. | tail -1 | awk '{ print sprintf("%.2f",$2/1073741824) }')" get_conf osd_keyring "$osd_data/keyring" "keyring"