Editing VPS Management (section)

== Misc jail Items ==

We are overselling hard drive space on jail2, jail8, jail9, a couple jails on jail17, jail4, jail12 and jail18.
Even though the vn file shows 4G size, it doesn’t actually occupy that amount of space on the disk. So be careful not to fill up drives where we’re overselling – use oversellcheck to confirm you’re not oversold by more than 10G.
There are other truncated jails, they are generally noted in a the file on the root system: /root/truncated

The act of moving a truncated vn to another system un-does the truncating- the truncated vn is filled with 0’s and it occupies physical disk space for which it’s configured. So, you should use dumpremote to preserve the truncation.

* if you are getting disk full messages for a BSD customer, it's fairly safe to clear out their /usr/ports/distfiles dir
* 4.x: ps and top an only be run by root in these jails.  Done on purpose:
As for `ps` and `top` - non-root users can run them, just not successfully because we have locked the permissions on /dev/mem and /dev/kmem to be root-readable only.  That is why non-root users cannot successfully run `ps` and `top`.
* user quotas do not work on freebsd jails - you cannot set up quotas at all, and that's that.
* You cannot inject a process into a 4.x jail, only HUP running processes. 6.x and onward you can with jexec
* jails see base machine’s uptime/load when running top/w
* if someone is unable to get in- cant ping, etc- see if they were blocked by castle (DoS), see if their ip is on the system (post reboot it was lost cause wasn’t in the rc.conf). preboot should catch that.
* in FreeBSD you cant su to root unless you belong to wheel group – so if you remove your acct and setup a new one, we have to add it to wheel group (add to /etc/group)
* Dmesg from underlying sys is seen in customers dmesg on jail
* Popper process going crazy @ 40% for 10min = someone who leaves mail on server
* Don’t force umounts on 4.x jails – it crashes the machine, generally ok on newer machines (running md)
* Good book for admin http://search.barnesandnoble.com/booksearch/isbnInquiry.asp?userid=t824VyRAYz&isbn=0596005164&itm=2
* Self-sign ssl cert http://httpd.apache.org/docs/2.0/ssl/ssl_faq.html#selfcert
* conversation with Glenn about semaphores and pgsql
<pre>SDBoody: hey, these are valid amounts/figures/increments right:
 kern.ipc.semmni=1280
 kern.ipc.semmns=1280
 
gr8feen: probably... I always forget exactly what those are, so I usually have to look them up
SDBoody: semaphores
SDBoody: i took the current 1024 and added 256 to them
SDBoody: need more for pgsql
gr8feen: I meant the mni and mns parts...
gr8feen: some of those are not ones you just want to add to...   hang on a sec and I'll look them up..
gr8feen: what's semmsl set to?
SDBoody: kern.ipc.semmsl: 1024
 
SDBoody: kern.ipc.msgseg: 2048 
 kern.ipc.msgssz: 8 
 kern.ipc.msgtql: 40 
 kern.ipc.msgmnb: 2048 
 kern.ipc.msgmni: 40 
 kern.ipc.msgmax: 16384 
 kern.ipc.semaem: 16384 
 kern.ipc.semvmx: 65534 
 kern.ipc.semusz: 152 
 kern.ipc.semume: 10 
 kern.ipc.semopm: 100 
 kern.ipc.semmsl: 1024 
 kern.ipc.semmnu: 512 
 kern.ipc.semmns: 1024 
 kern.ipc.semmni: 1024 
 kern.ipc.semmap: 768 
 kern.ipc.shm_allow_removed: 0 
 kern.ipc.shm_use_phys: 1 
 kern.ipc.shmall: 262144 
 kern.ipc.shmseg: 256 
 kern.ipc.shmmni: 784 
 kern.ipc.shmmin: 1 
 kern.ipc.shmmax: 536870912 
 kern.ipc.maxsockets: 25600 
 
gr8feen: ok...msl is max per id, mni is max ids, mns is max number of semaphores... so you probably want something like mns = mni * msl
gr8feen: which one did you run out of?
SDBoody: not sure how to tell- ipcs shows the sems in use add up to 1024
SDBoody: there are 59 entries
gr8feen: I'm assuming you tried to start postgres and it failed?
SDBoody: yes
gr8feen: it should have logged why, somewhere..
gr8feen: if I recall, it'll tell you which one it ran out of
SDBoody: > DETAIL:  Failed system call was semget(1, 17, 03600).
 
gr8feen: so it wanted an id with 17 semaphores...I'd start by making mns = 17*mni and leave mni and mnl set to what they are now and see what it does
SDBoody: i think mni is plenty high
SDBoody: ok, more reasonable 17408
gr8feen: yeah...just change that one and see how it goes..
SDBoody: and leave mni alone at 1024?
gr8feen: yeah...mni id the max number of ids...but if it's trying to get something like 17 per id, your going to hit mns before you hit anything else
SDBoody: right, but doesn't hurt to have it that high (assuming)

gr8feen: not really...  I think those get allocated out of ram that you cant page out, but it's still such a small amount that it really doesn't matter
SDBoody: looks like that worked, thx!
gr8feen: cool

see semaphores: ipcs -a -s</pre>