Ben Williams's questions

I have a process - a perl script - that does:

while true
    check a POP account on a server on the lan
    process any email found
    write logs - messages found, actions taken, errors
    sleep for 15 seconds

It's running on a redhat 7.3 server (I inherited it, I'm not happy about the age of that box). It's run out of /etc/inittab like:

spop:2345:respawn:/usr/local/gw/bin/popdmn 

If it dies, init restarts it.

In the last couple of days, the process will no longer work unless it's straced. When it's just running, it never logs into the pop server. As soon as it's straced (via "strace -Ff -p cat /usr/local/gw/var/popdmn.pid"), it works flawlessly.

As a workaround, I'm running screen on the server with an strace running. Obviously this is less than ideal.

Why would a process do this? I haven't seen this happen before.

Is there a way to tell the format of an svn repo on disk, without knowing how it was created? Possibly from the format or db/format files?

This question has lots of useful info about the different contents of those files, but there's no answer to what backend format the repo is using.

The (gentoo) machine I'm on is running subversion 1.5.6:

me@foo ~ $ svn --version
svn, version 1.5.6 (r36142)
   compiled Jan 13 2010, 15:02:26

I created a couple of test repos to play with:

me@foo ~ $ svnadmin create --fs-type bdb bdb-repo
me@foo ~ $ svnadmin create --fs-type fsfs fsfs-repo

The format and db/format files look almost identical:

me@foo ~ $ more bdb-repo/format bdb-repo/db/format
::::::::::::::
bdb-repo/format
::::::::::::::
5
::::::::::::::
bdb-repo/db/format
::::::::::::::
3
me@foo ~ $ more fsfs-repo/format fsfs-repo/db/format
::::::::::::::
fsfs-repo/format
::::::::::::::
5
::::::::::::::
fsfs-repo/db/format
::::::::::::::
3
layout sharded 1000

The only difference being the layout sharded 1000 line in the fsfs repo's db/format file. Is that the definitive indicator that it's using fsfs format? If not, is there some other way to tell?

At my workplace, we're using sudo-ldap to provide sudo privileges across the network from a centralised store. We do this based on the username plus the hostname they're running sudo on, e.g.:

sudoHost: foo1
sudoUser: user1

I want to be able to match the sudoHost attribute based on the FQDN rather than the short hostname, so that we can use DNS subdomains for different environments - e.g. use web1.dev.example.com in the development environment and web1.test.example.com in the test environment and so on. However, right now sudoHost only matches on the short hostname.

i.e. this works:

sudoHost: foo1

But this doesn't work:

sudoHost: foo1.dev.example.com

The problem with just having the short hostname (foo1) is that it matches across all environments - foo1.dev, foo1.test, foo1.staging, etc. I want it to be more granular than that.

Is it possible to get sudo-ldap to query based on the FQDN (as returned by hostname -f) rather than just the short hostname?

I've googled with no success. I can't find anything in the man page or online; the closest it gets when talking about sudoHost attribute is:

sudoHost

A host name, IP address, IP network, or host netgroup (prefixed with a '+'). The special value ALL will match any host.

No mention of FQDN vs. short hostname. There was no mention of FQDN vs. short hostname in the Configuring ldap.conf section either.

The host in question is running Ubuntu 10.04.

We've got seperate environments at my workplace for development, testing, integration, and staging.

Within those envs, we've overloaded the hostnames in DNS - e.g. in the dev environment, the primary web machine is called web1.dev.example.com, and in the test environment, the primary web machine is web1.test.example.com.

To distinguish between machines in the different environments, I want to customise the bash prompts to display the FQDN rather than just the hostname. Well and good; I should be able to replace \h with \H in $PS1, right? Hmm. They show the exact same thing.

me@web1:~$ hostname
web1
me@web1:~$ hostname -f
web1.dev.example.com
me@web1:~$ export PS1="\[\u@\h: \w\]\$ "
me@web1: ~$ export PS1="\[\u@\H: \w\]\$ "
me@web1: ~$ 

In /etc/hostname, I've got just the hostname (web1). hostname and hostname -f both return the correct results ("web1" and "web1.test.example.com" respectively), and I've got the correct entries in /etc/hosts.

What gives?

These are Ubuntu 10.04 hosts, if that makes a difference.

I'm having a problem with the raid array in a server (Ubuntu 10.04).

I've got a raid5 array of 4 disks - sd[cdef], created like this:

# partition disks
parted /dev/sdc mklabel gpt
parted /dev/sdc mkpart primary ext2 1 2000GB
parted /dev/sdc set 1 raid on
# create array
mdadm --create -v --level=raid5 /dev/md2 /dev/sdc1 /dev/sdd1 /dev/sde1 /dev/sdf1

This has been running fine for a couple of months.

I just applied system updates and rebooted, and the raid5 - /dev/md2 - didn't come back on boot. When I re-assembled it with mdadm --assemble --scan, it seems to have come up with only 3 of the member drives - sdf1 is missing. Here's what I can find:

(Side-note: md0 & md1 are raid-1 built on a couple of drives, for / and swap respectively.)

root@dwight:~# mdadm --query --detail /dev/md2
/dev/md2:
        Version : 00.90
  Creation Time : Sun Feb 20 23:52:28 2011
     Raid Level : raid5
     Array Size : 5860540224 (5589.05 GiB 6001.19 GB)
  Used Dev Size : 1953513408 (1863.02 GiB 2000.40 GB)
   Raid Devices : 4
  Total Devices : 3
Preferred Minor : 2
    Persistence : Superblock is persistent

    Update Time : Fri Apr  8 22:10:38 2011
          State : clean, degraded
 Active Devices : 3
Working Devices : 3
 Failed Devices : 0
  Spare Devices : 0

         Layout : left-symmetric
     Chunk Size : 64K

           UUID : 1bb282b6:fe549071:3bf6c10c:6278edbc (local to host dwight)
         Events : 0.140

    Number   Major   Minor   RaidDevice State
       0       8       33        0      active sync   /dev/sdc1
       1       8       49        1      active sync   /dev/sdd1
       2       8       65        2      active sync   /dev/sde1
       3       0        0        3      removed

(Yes, the server's called Dwight; I'm a The Office fan :) )

So it thinks one drive (partition really) is missing, /dev/sdf1.

root@dwight:~# mdadm --detail --scan
ARRAY /dev/md0 level=raid1 num-devices=2 metadata=00.90 UUID=c7dbadaa:7762dbf7:beb6b904:6d3aed07
ARRAY /dev/md1 level=raid1 num-devices=2 metadata=00.90 UUID=1784e912:d84242db:3bf6c10c:6278edbc
mdadm: md device /dev/md/d2 does not appear to be active.
ARRAY /dev/md2 level=raid5 num-devices=4 metadata=00.90 UUID=1bb282b6:fe549071:3bf6c10c:6278edbc

What, what, /dev/md/d2? What's /dev/md/d2? I didn't create that.

root@dwight:~# cat /proc/mdstat
Personalities : [linear] [multipath] [raid0] [raid1] [raid6] [raid5] [raid4] [raid10] 
md2 : active raid5 sdc1[0] sde1[2] sdd1[1]
      5860540224 blocks level 5, 64k chunk, algorithm 2 [4/3] [UUU_]

md_d2 : inactive sdf1[3](S)
      1953513408 blocks

md1 : active raid1 sdb2[1] sda2[0]
      18657728 blocks [2/2] [UU]

md0 : active raid1 sdb1[1] sda1[0]
      469725120 blocks [2/2] [UU]

unused devices: <none>

Ditto. md_d2? sd[cde]1 Are in md2 properly, but sdf1 is missing (and seems to think it should be an array of its own?)

root@dwight:~# mdadm -v --examine /dev/sdf1
/dev/sdf1:
          Magic : a92b4efc
        Version : 00.90.00
           UUID : 1bb282b6:fe549071:3bf6c10c:6278edbc (local to host dwight)
  Creation Time : Sun Feb 20 23:52:28 2011
     Raid Level : raid5
  Used Dev Size : 1953513408 (1863.02 GiB 2000.40 GB)
     Array Size : 5860540224 (5589.05 GiB 6001.19 GB)
   Raid Devices : 4
  Total Devices : 4
Preferred Minor : 2

    Update Time : Fri Apr  8 21:40:42 2011
          State : clean
 Active Devices : 4
Working Devices : 4
 Failed Devices : 0
  Spare Devices : 0
       Checksum : 71136469 - correct
         Events : 114

         Layout : left-symmetric
     Chunk Size : 64K

      Number   Major   Minor   RaidDevice State
this     3       8       81        3      active sync   /dev/sdf1

   0     0       8       33        0      active sync   /dev/sdc1
   1     1       8       49        1      active sync   /dev/sdd1
   2     2       8       65        2      active sync   /dev/sde1
   3     3       8       81        3      active sync   /dev/sdf1

...so sdf1 thinks it's part of the md2 device, is that right?

When I run that on /dev/sdc1, I get:

root@dwight:~# mdadm -v --examine /dev/sdc1
/dev/sdc1:
          Magic : a92b4efc
        Version : 00.90.00
           UUID : 1bb282b6:fe549071:3bf6c10c:6278edbc (local to host dwight)
  Creation Time : Sun Feb 20 23:52:28 2011
     Raid Level : raid5
  Used Dev Size : 1953513408 (1863.02 GiB 2000.40 GB)
     Array Size : 5860540224 (5589.05 GiB 6001.19 GB)
   Raid Devices : 4
  Total Devices : 3
Preferred Minor : 2

    Update Time : Fri Apr  8 22:50:03 2011
          State : clean
 Active Devices : 3
Working Devices : 3
 Failed Devices : 1
  Spare Devices : 0
       Checksum : 71137458 - correct
         Events : 144

         Layout : left-symmetric
     Chunk Size : 64K

      Number   Major   Minor   RaidDevice State
this     0       8       33        0      active sync   /dev/sdc1

   0     0       8       33        0      active sync   /dev/sdc1
   1     1       8       49        1      active sync   /dev/sdd1
   2     2       8       65        2      active sync   /dev/sde1
   3     3       0        0        3      faulty removed

And when I try to add sdf1 back into the /dev/md2 array, I get a busy error:

root@dwight:~# mdadm --add /dev/md2 /dev/sdf1
mdadm: Cannot open /dev/sdf1: Device or resource busy

Help! How can I add sdf1 back into the md2 array?

Thanks,

• Ben

I can't run crontab -e as a regular user on a gentoo box I've got. Here's a transcript:

ben@tools ~ $ crontab -e
    # (I add something trivial, like a new line containing "#foo")
crontab: installing new crontab
chown: Operation not permitted
crontab: edits left in /tmp/crontab.XXXXULPy6M
ben@tools ~ $

I have another gentoo machine on which it works fine.

I've checked permissions on all these things, and they all look both sensible and identical to the machine where it works:

tools ~ # ls -AFlh /usr/bin/crontab 
-rwxr-s--x 1 root crontab 29K Oct 22 16:16 /usr/bin/crontab*
tools ~ # ls -AFlhd /tmp
drwxrwxrwt 17 root root 3.2K Oct 22 16:34 /tmp/
tools ~ # ls -AFlhd /var/spool/cron
drwxr-xr-x 4 cron root 4.0K Oct 22 16:07 /var/spool/cron/
tools ~ # ls -AFlhd /var/spool/cron/crontabs
drwx-wx--T 2 root crontab 4.0K Oct 22 16:16 /var/spool/cron/crontabs/

I am a member of the groups cron and crontab:

ben@tools ~ $ groups ben
cron crontab sysops

there is no /etc/cron.allow, and /etc/cron.deny is blank apart from comment lines.

I've tried updating the vixie-cron package, by doing:

emerge --sync
emerge -v vixie-cron # reinstall the current version, in case something is wrong
# add "sys-process/vixie-cron ~x86" to /etc/portage/package.keywords
emerge -v =vixie-cron-4.1-r11

No good.

I've tried running strace against it (both as myself with strace -Ff crontab -e and as root by running crontab -e as myself and then strace -FF -p foo as root), but I can't see anything more useful than the error message I've already got:

chown32("crontabs/tmp.XXXXI5Oi61", 0, -1) = -1 EPERM (Operation not permitted)
write(2, "chown: Operation not permitted\n", 31) = 31
close(4)                                = 0

I have of course googled for it, and every google result has talked about checking permissions on the same files/folders I've checked above.

Help! I'm out of ideas.

• Ben

Is anyone running Zend Server under Daemontools?

I know I can use -D NO_DETACH to run apache under daemontools. But, I'm not sure if the other Zend Server components (monitor, lighthttpd, scd, jobqueue) have similar options.

(The end goal is "run Zend Server with service supervision", so upstart would be fine too. For reference, my servers are running Ubuntu 10.04 LTS.)

• Ben