Finding who is using dovecot’s imap process in the process table

Ever want to know who has what imap-login ? Then run something along these lines

for i in $(ps -ef | awk '/[i]map$/ {print $2}'); do
      echo "$(lsof -p $i | awk '/cwd/ {print $9}') is handled by process id $i" ;
done

Which returns something along these lines (names changed to protect the innocent ;) .

$PATHTOVPOPMAIL/domains/$DOMAIN/$USER is handled by process id 1277

Just a quick lunch time hack that might help some errant google searcher.

Getting X to work with 82945G/GZ Integrated Graphics Controller

This is a easy one to fix you just have to open:

/etc/grub.conf

and find

nomodeset

remove it,then move your xorg.conf (if you made one trying to fix this mess) outta the way, then reboot.

Be Quiet PHP (5.4)!

Well thought I would drop in and writeup a quick fix to a minor annoyance with the IUS (and other 5.4 versions of php) and that is the STRICT and Deprecated warnings in logs. It appears that the new version of this is a bit more “open” on what it takes to turn this on (ie code in sites like phpbb, wordpress, and others) and the error_reporting setting in php.ini didnt make it quiet. Lucky for me this quick hack did:

emacs /etc/httpd/conf.d/php.conf

<IfModule prefork.c>
LoadModule php5_module modules/libphp5.so
php_admin_value error_reporting ” E_ALL & ~E_DEPRECATED & ~E_STRICT”
</IfModule>

Global override saves the day. Now I can go back to sed-ing out all this cruft this update to 5.4IUS on RHEL 6 just did to my logs! I will say that the mysqlnd that comes with 5.4 is quite impressive…look for a writeup sometime soon.

Ghost in the machine ?!

Here is a real error I got after making a system config change recently. Can anyone guess what that change might have been. To give a little hint I got this on a new session opened up in screen on the system that had the change.

[I have no name!@*****]>ssh ******
You don’t exist, go away!

How To find the model of a USB harddrive in a USB enclosure in Linux

NOTE: This is not working totally with USB3 interfaced drives. Likely a update to the tools will fix this.

The following command:

smartctl -i /dev/sdb

Where hdparm /sdparm (and lsusb -v) show you info about the USB device this gives you the info about the device IN the USB enclosure.
Here is output:
=== START OF INFORMATION SECTION ===
Model Family:     Seagate Barracuda XT
Device Model:     ST33000651AS
Serial Number:    9XK0AKRX
LU WWN Device Id: 5 000c50 02d957ba3
Firmware Version: CC43
User Capacity:    3,000,592,982,016 bytes [3.00 TB]
Sector Size:      512 bytes logical/physical
Device is:        In smartctl database [for details use: -P show]
ATA Version is:   8
ATA Standard is:  ATA-8-ACS revision 4
Local Time is:    Fri Apr 12 08:17:27 2013 EDT
SMART support is: Available – device has SMART capability.
SMART support is: Enabled

Getting NBD (Network Block Device) Back in RHEL 7.x, CentOS 7.x, RHEL 6.x and CentOS 6.x

The quick and dirty way:

  1. Download SRC rpm for kernel you want to build module for
  2. Run `rpmbuild -bp kernel.spec`. While the magic is happening here go `make pbnj` or read the latest copy of 2600…oh wait thats not printed any more..crap.
  3. Next go to your your kernel build directory `cd /root/rpmbuild/BUILD/kernel-$(uname -r | sed ‘s/.x86_64//’)/linux-$(uname -r)`
  4. type `make menuconfig`
  5. Go to “Device Drivers | Block devices” and set “M” on “Network block device support” exit saving new kernel config.
  6. Next type `make prepare && make modules_prepare && make` (make is needed otherwise you are missing symbols that the nbd.ko needs)
  7. If your system is older, now is the time to take your significant other to dinner, catch up on some online gaming,  or play with your kids.
  8. Next we build just the block device modules by typing`make M=drivers/block`
  9. We now check the module ` modinfo drivers/block/nbd.ko` if all looks good we copy it to the running kernels extra folder in /lib/modules/$(uname -r)/extra.
  10. We can finally now load the module via `depmod -a && modprobe nbd`.

I will be doing some testing on NBD vs GlusterFS in a coming post. The one kicker is that the NBD and GlusterFS servers will be running on a RazPi’s with a 2G USB thumbdrive for the exported block device (thanks to the vendors at conferences!) .

Quick Hack to email CPU lmsensors thresholds

Below is a quick hack I did that I thought folks might find useful. I used it to watch the CPU temps of a system in a VERY warm room. More a exercise but someone might find it useful :)

#!/bin/sh
EMAIL=YOUREMAILHERE
SENSOR=coretemp-isa-0000
MAXTEMP=50
MESSAGE=/dev/null
HOSTNAME=$(hostname)

for t in $(sensors -u $SENSOR | grep input | awk -F: ‘{printf “% .0f”, $2}’); do
if [ $t -gt $MAXTEMP ] ;
then mail -s “$(echo “CPU in $HOSTNAME at “$t”C”)” $EMAIL < $MESSAGE;
fi;
done

How (you might be able) to save a archive from the dreaded “premature end of file” !

Unless the archive is totally foobar one way is to use this quick command

pax -v -r * <  /path/to/archive

This will restore your archive to the paths in the archive (ie /home/user will be restored to that path) if you need different options man pax is your new friend.

Linux CLI on the FLY (interface stats)

I am going to start a new section of Linux CLI foo  that you may or may not know about. Many of these will be faster ways of doing things or getting information in a slightly different fashion. Many are old favorites of mine that I just wanted to share in case they help someone else.

Now without further ado the first command is :

ip -s l

short for ip stats list this gives a fast overview of your interfaces and  the types of packets that have been going through them. Output looks like this :

[root@pbnj lisa11]# ip -s l
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
RX: bytes  packets  errors  dropped overrun mcast
1084866683 400954   0       0       0       0
TX: bytes  packets  errors  dropped carrier collsns
1084866683 400954   0       0       0       0
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000
link/ether 00:30:48:61:c7:64 brd ff:ff:ff:ff:ff:ff
RX: bytes  packets  errors  dropped overrun mcast
552827144  1541759  0       0       0       0
TX: bytes  packets  errors  dropped carrier collsns
1428902993 1445402  0       0       0       0
3: eth1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop qlen 1000
link/ether 00:30:48:61:c7:65 brd ff:ff:ff:ff:ff:ff
RX: bytes  packets  errors  dropped overrun mcast
0          0        0       0       0       0
TX: bytes  packets  errors  dropped carrier collsns
0          0        0       0       0       0

More information on interface stats can be found via:
ethtool -S $DEVICE
ifconfig $device (or just ifconfig for all devices)
ifconfig -s $DEVICE

among others….

Sluggish LVM/Disk/Raid performance under Linux (RHEL/CentOS 6)? Check power management!

If you are running tuned these settings will likely “undo” themselves. I found this out the hardway ;)

Update: If you are running a old firmware (older than CC34) update !! More on the update are on Seagates page and here http://niallbest.com/seagate-2tb-st32000542as-cc35-firmware-upgrade/. To find your firmware use `hdparm -I /dev/sd[a-z] | grep Firmware`.

I was noticing Samba being slow when I would access it from time to time. It would always work but man was it annoying. Since this was right after I added some new 2TB drives to my LVM/Software Raid 1 array I figured I would check to see if power management had been the cause for my pain. Sure enough I found this :

[root@stardust ~]# hdparm -C /dev/sd[a-z] | sed ‘/^$/d’
/dev/sda:
drive state is: active/idle
/dev/sdb:
drive state is: standby
/dev/sdc:
drive state is: active/idle
/dev/sdd:
drive state is: active/idle
/dev/sde:
drive state is: standby
/dev/sdf:
drive state is: active/idle
/dev/sdg:
drive state is: active/idle

To which I answered with :

hdparm -B 255 -S 0 -K 1 /dev/sd[b,e]

and wala all was happy again and my delay was gone!

/dev/sda:
drive state is: active/idle
/dev/sdb:
drive state is: active/idle
/dev/sdc:
drive state is: active/idle
/dev/sdd:
drive state is: active/idle
/dev/sde:
drive state is: active/idle
/dev/sdf:
drive state is: active/idle
/dev/sdg:
drive state is: active/idle

The new drives were Seagate “Green” drives (ST32000542AS) so I cant blame them for shipping with aggressive power save settings. If you need to check what your power save settings are you can do something like :

for d in /dev/sd[a-z]; do echo $d; hdparm -I $d | grep level; done

/dev/sda
/dev/sdb
Advanced power management level:  disabled
/dev/sdc
/dev/sdd
/dev/sde
Advanced power management level: disabled
/dev/sdf
/dev/sdg
HDIO_DRIVE_CMD(identify) failed: Invalid exchange

The /dev/sdg device is a USB drive and they dont play nice with hdparms commands usually.