After I thought I understood memory manager after all, this comes along. I have a behavior I can’t understand or explain. Let’s see if anyone can solve the puzzle.
You have a machine, lets say with 8GB Ram. You only run a small number of process on that machine and you want to add another service.
You run free -m and it looks like this:
total used free shared buffers cached
Mem: 7980 4814 3165 0 1 4593
-/+ buffers/cache: 220 7759
Swap: 1759 0 1759
And you think hey, no problem a lot of free memory available. 220MB “real used” memory and 7759MB “freeable memory”. Let’s bring it on.
You start you application (I use memhog from numactl) which eats 4G of RAM and gives you a OOM.
Who knows why?
(The solution is attached as base64 encoded block) To get a peek simple execute the block in a shell.
cat << EOF | openssl base64 -d
I would still love an explanation why it is that way.
Be careful if you update to RHEL6.4 and you are using NX. RHEL6.4 removes the keymap.dir and this breaks NX at the moment.
To fix this, simply do a touch /usr/share/X11/xkb/keymap.dir. It looks like that RedHat will not fix this, because the file layout of xkeyboard-config is not part of the fixed ABI/API. The Centos guys are aware of this, and I hope they will fix this instead.
Even with the new fedup tool to upgrade. There are a couple of commands I use always after an upgrade. I suggest you use them too:
- yum distro-sync
- package-cleanup –problems
- package-cleanup –orphans
- rpmconf -a -fvimdiff
Helps to keep your Fedora installation clean.
After joining my local LUG at the python workshop and because I start using more python now. Here some hints and tools that I’m using a lot lately.
Install it with pip install virualenvwrapper or yum install python-virtualenvwrapper.noarch.
Create new Enviroments with: mkvirtualenv ENV
Jump into the Env: workon ENV
jump out of Env: deactivate
If you are using ArgoUML and thought “Hey, its written with Eclipse, lets try using it inside Eclipse”. I’m pretty sure you search found ArogEclipse.
Please be aware this project looks very dead. I tried to open an perfectly good zargo file with it and it tells me it is broken. The Version is from 2010. Please use the standalone Version ArgoUML!
I setup a RedHat Enterprise Box that should serve as a httpd server for a Suse Enterprise Server zypper repository and it don’t work.
Simple curl also don’t work. If you use curl -1 or curl -3 it works like a charm, curl without option I get the following error message:
Error code: Unrecognized error
error message: error: 14077458:SSL routines:SSL23_GET_SERVER_HELLO:reason(1112)
After some wiresharking I found the solution. You have to set a ServerName in the ssl VirtualHost. Yes, I know the hostname/CommonNamem is already in the sslcert you are using, but you have to repeat it in the config file there, otherwise some older version of openssl (like the one from SLES11) will run into this kind of problems.
I bought an IPhone for my mum this morning and had an interesting experience. Firefox on Mac showed the shop with a “Please wait” sign, and with Safari the shop was open and ready to order. I Pressed Ctrl+Shift+r for real reload, but still “Please wait”. Both on the same machine. Did someone had the same experience, or was that simple bad luck?
I have seen a lot of broken and bad RPM packages.
From vmware packages that contains one big tarfile and the rpm-postscript uncompressed this into a directory.
To test scripts from Oracle DB 11.2 that does not detect the 32bit libraries from Oracle Linux 6.3 because it looks for i386 instead of i686.
I get used to that. But Hitachi just shot the hoop.
You can not install the Hitachi Software if the rpm is not in /media/cdrom/rpms/,the postscript is over 10000 lines and the x86_64 packages requires 32bit libraries.
I start to think that Enterprise Software is an alias for broken packages and bad software installations.
We searched for multiple days why one of our cluster machines was hanging for a couple of seconds, before it was fenced. We found out that the iptables LOG did put to much output onto the serial.
I wrote a kernel module to create a hugh amount of messages, and the serial line got scrambled. I got this output on the serial:
[ 222.111488] Clocksource tsc unstable (delta = -60129537542 ns). Enable clocksource lover by adding clocksource_failover kernel parameter.
but this in the syslog:
[ 222.111488] Clocksource tsc unstable (delta = -60129537542 ns). Enable clocksource failover by adding clocksource_failover kernel parameter.
The romatic guy in me wants to keep the serial version 🙂
As a lot of people have noticed Oracle is offering a “conversion” script to move from Centos to Oracle Linux. Of course it is not really a conversion, it only moves the yum repos from Centos to Oracle. The existing packages are not touched. If you want a complete conversion have a look at my page about moving from Centos to Scientific Linux. You could use the same mechanism to move from any RHEL-based system to any other RHEL-based system.
I checked the announcement mails for the last kernel update RHSA-2012:1156-01. Here the recieved time for the mail on my mailserver.
RedHat: 14 Aug 2012 20:25:42 +0200 (CEST)
Centos: 15 Aug 2012 06:26:25 +0200 (CEST)
SL: 15 Aug 2012 16:36:55 +0200 (CEST)
Oracle: 15 Aug 2012 22:10:25 +0200 (CEST)
With this information in mind this Graph looks “interesting”.
But overall I like the idea that nowadays we are talking about hours of delay instead of days or weeks between Red Hat and the Clones.