.powerdirector discount

So I’m speccing out some new hardware at work and .powerdirector discount have had to dredge up a bunch of dusty old RAID knowledge, while also getting familiar with LSI’s MegaCLI tools which are.. lets just say under-documented. First off, you .powerdirector discount should keep the Wikipedia page for RAID handy as a reference, it .powerdirector discount does a good job of explaining and visualizing all the different raid levels.

Standard RAID Levels and Nested RAID Levels

Nested RAID is .powerdirector discount basically multiple Standard RAID Levels that are spanned creating what amounts to .powerdirector discount a meta-RAID. This is done to improve performance, redundancy or both. The two main aspects of performance and .powerdirector discount redundancy are striping and mirroring, respectively. Mirroring is just what it .powerdirector discount sounds like, each drive is an exact mirror of the others. This results in redundancy but you’ll generally have .powerdirector discount slower write times. Striping is where sequential sectors are written and .powerdirector discount read across multiple disks in the array, giving faster seek times and .powerdirector discount general performance but without redundancy. There is also parity which is a bit more complicated but gives distributed redundancy. RAID 5 is an example of this.

So that’s some very basic RAID information. The rest of this .powerdirector discount is just a little cheat sheet for the MegaCli command line:

Linux Problems lsi, megacli, raid

So this isn’t your normal hanging directory listing; it’s not caused by a .powerdirector discount huge volume of files or by broken NFS or other ‘legitimate’ issues. If you’ve reached this step in your online searching, chances are you’ve ruled out all the .powerdirector discount normal causes for a hanging directory listing. So what this might be .powerdirector discount for you is a combination of the following factors:

• ls -l –color=tty (colorized ls output)
• symlinks to unavailable resource
• autofs

Specifically what .powerdirector discount is happening is there is a directory with symlinks, which point to .powerdirector discount NFS resources controlled by autofs. However, autofs only actively mounts a .powerdirector discount resource when it is requested. And the colorizing code portion of ls won’t colorize the .powerdirector discount output until it has assessed what type of resource is being handled. So the .powerdirector discount flow goes something like this:

1. User requests ls with the –color=tty option set (often aliased to ls)
2. ls hits the first symlink and tries to classify its target type
3. The autofs resource is .powerdirector discount not mounted or triggered by ls trying to classify it
4. This will now hang, sometimes indefinitely

You can verify this is the cause by three separate methods:

• ls -l –color=never /path
• strace ls /path
• ls -l /path/to/symlink/target

The first one simply removes the .powerdirector discount color coding from the ls request and the second also ignores it .powerdirector discount as strace handles the stdout. Finally, the last one simply avoids the .powerdirector discount color coding + symlink to autofs all together. This also (hopefully) shows that .powerdirector discount autofs is actually properly mounting the NFS resource. If the last test fails, then .powerdirector discount you have an issue with autofs.

It’s kind of a weird bug, but there you go.

Linux Problems autofs, color coding, ls hangs, symlink

If you .powerdirector discount are getting this error printing to the console or in your .powerdirector discount logs, chances are your route table is getting filled up.  First thing you .powerdirector discount want to do is determine what your max route cache size is:

sysctl -a |grep route.max_size

This should be .powerdirector discount a fairly high number. To contrast it against how many route cache entries you .powerdirector discount have, use:

ip route show cache|grep -c via

If you want to clear the route cache, you can run:

ip route flush cache

Personally I’ve most often see this .powerdirector discount behavior when a server is getting a high volume SYN flood attack. So the .powerdirector discount obvious thing to do would be to mitigate the SYN flood. If all you .powerdirector discount have available is iptables, you might end up crashing the .powerdirector discount server due to resources, as custom filters for high volume/small size attacks like SYN floods will chew up CPU. So realistically you .powerdirector discount should be filtering it at a firewall or some other sort of network protection.

Failing that, or if it’s going to .powerdirector discount take awhile to get that setup, you can run a while true loop to .powerdirector discount clear the route table cache every few seconds:

while true; do ip route flush cache; sleep 5; done &

That’s a bit of a hack, but I’ve had .powerdirector discount it keep servers up and running and serving client sites through Apache despite a .powerdirector discount 33kpps SYN flood.

Enjoy!

Linux Problems dst cache overflow, route cache bd encryption

Wow, what a blog-slacker I have been. The reality is that I’ve been .powerdirector discount just too busy at work to make time for any continued posting. Right now I am .powerdirector discount working on a large-scale (hundreds of servers, thousands of containers) migration from .powerdirector discount one virtualized environment to another. This is a significant undertaking with a .powerdirector discount lot of work having been put into it but it will definitely provide a .powerdirector discount far superior customer experience for our clients, as well as make administrating that .powerdirector discount environment much easier.

I’ve also been .powerdirector discount coding a lot of tools and working with our IT security department on a .powerdirector discount special project, which is about all I can say about that. As for .powerdirector discount tech tips.. hmm… nothing too amazing, but if you’re having problems with mixed local/nfs content and .powerdirector discount you want to be sure you can safely remove only local content from .powerdirector discount a tangled tree of sub-directories.. don’t forget one of your best friends, ‘find’.

find /path -type f -mount

That will show only the local files, as it won’t descend into other file systems. Optionally you .powerdirector discount can check the results with something like:

find . -printf “%F %p\n”

That will show the .powerdirector discount file system type, then the full path. Local content should be .powerdirector discount ‘ext2′,’ext3′ and so on. Nas mounts should print out ‘nfs’.

Okay that’s all I have time for.. I’m off to .powerdirector discount replace a burned out headlight in one of my cars, what .powerdirector discount fun!

Everything Else

Okay so I’ve been .powerdirector discount gone for quite awhile. I blame work and a lot of crazy busy life stuff, including a .powerdirector discount newly discovered love for fitness. Running, hiking, biking, yoga, weights, the .powerdirector discount works. It’s horribly time consuming!

Anyway, I do have .powerdirector discount something techy to share, so lets get right to it.

ERROR: mount: Unable to bindresvport TCP socket: errno 98 (Address already in use)

You may run into this .powerdirector discount when troubleshooting an NFS problem. In my case I was trying to .powerdirector discount figure out why trying to pull a directory listing on the .powerdirector discount NFS mount was returning IO errors:

ls /path/to/nfs_mount
ls: /path/to/nfs_mount: Input/output error

Of course the .powerdirector discount first thing I did was try to unmount it and remount it. This is .powerdirector discount when I got the address already in use error. This included the .powerdirector discount ports 2049 and 4046 in the error output:

mount: trying 1.2.3.4 prog 100003 vers 3 prot tcp port 2049
mount: trying 1.2.3.4 prog 100005 vers 3 prot tcp port 4046

Naturally I looked at what .powerdirector discount was using those ports using lsof and netstat. I found httpd and .powerdirector discount rpc.mount proceses and killed them, but the error persisted. Next I used rpcinfo, which showed rquotad was using 4046. A simple pkill -f rquotad and .powerdirector discount one more mount attempt cleared up the issue.

Like any other resource conflict, always look for .powerdirector discount the offender that is holding the resource you need and generally kill it. A reboot will usually do the .powerdirector discount same thing, but that’s a .powerdirector discount bit of a blunt force resolution. Finess will often save you .powerdirector discount downtime! To re-cap, here are .powerdirector discount some tools to use when finding what is using a resource:

• ps - aux, auxe, whatever your .powerdirector discount preferred flavor is.. verbosity is good!
• netstat -antp (my personal choice)
• lsof (great for files, paths and ports with -i:N)
• rpcinfo -p
• strace (dont’ forget to strace the failing service/process, you may track down the conflict that way)

And that’s pretty much it folks!

Linux Problems errno 98, mount, nfs, rquotad

Okay, so tonight’s lesson is .powerdirector discount about about process states and what to do when they become uninterruptible. First, stolen shamelessly from .powerdirector discount the man page:

PROCESS STATE CODES
Here are .powerdirector discount the different values that the s, stat and state output specifiers (header “STAT” or “S”) will display to describe the state of a process.

D Uninterruptible sleep (usually IO)
R Running or runnable (on run queue)
S Interruptible sleep (waiting for an event to complete)
T Stopped, either by a job control signal or because it is being traced.
W paging (not valid since the 2.6.xx kernel)
X dead (should never be seen)
Z Defunct (”zombie”) process, terminated but not reaped by its parent.

The main one we’re focusing on tonight is .powerdirector discount that first one, state D. Most of the time the rest of the .powerdirector discount states are only of passing interest when looking at a process list. Any time you .powerdirector discount see processes in state D though, it usually means a problem, and .powerdirector discount as the man page helpfully points out, it usually means an .powerdirector discount IO issue. Examples would be a full disk or a missing/down network filesystem.

Just to .powerdirector discount get it out of the way, sometimes the only way to .powerdirector discount clear these is to reboot the server. It sucks, but thems the .powerdirector discount breaks! However there are a few things to try first.

• Identify the .powerdirector discount process that is hanging. What does it do, what resources does it .powerdirector discount use that might be causing it to become unresponsive?
• Are all of your .powerdirector discount current mounts correct and responsive, both local and network? Any full disks?
• If possible re-run the .powerdirector discount process through an strace and see which resource call it .powerdirector discount gets stuck at.
• Use the .powerdirector discount output of the strace to determine the point of failure and .powerdirector discount try to go fix it!

Linux Problems ps, state D, strace, uninterruptible sleep

So this .powerdirector discount is a bit late, but all you folks out there really should make sure you .powerdirector discount have patched your Windows boxes against Conflicker. If you’re not sure if you .powerdirector discount have, check to see if KB958644 has been installed. From a .powerdirector discount command prompt: (note: if WMIC is .powerdirector discount not installed, it will do so when you run this, fear not, it .powerdirector discount is a legitimate Microsoft application)

wmic qfe | findstr “KB958644″

If this returns nothing, then you don’t have .powerdirector discount it installed. Otherwise it should show your workstation name and some brief info about the .powerdirector discount patch. If it’s not installed and .powerdirector discount Windows Update is not prompting with new downloads, you can force it .powerdirector discount to contact the WSUS servers by going to start, run and .powerdirector discount executing the following: (it has no display output FYI)

wuauclt /detectnow

Optionally, crack open your .powerdirector discount control panel and launch Automatic Updates. Even if you don’t have the patch yet, it’s not too late to apply it. Personally I don’t let Windows Update run on auto-pilot, but that .powerdirector discount is just because I’m a control freak and like to do it on my own schedule.

Last bit about Windows Updates, you can review the full log, it’s located at %windir%\system32\WindowsUpdate.log, which should be .powerdirector discount a configured as a viable system path, so you can .powerdirector discount just go to start, run and type in WindowsUpdate.log and .powerdirector discount it will come up. Enjoy!

Windows Problems Conflicker, KB958644, Windows Update serial for vegas pro 10 64 bit, WSUS

Another big gap between posts; c’est la vie! Anyway, on to the tech.

Error: semget: No space left on device

You may come across this .powerdirector discount from time to time, usually on a pretty busy web server. In my case it .powerdirector discount showed up in the Apache error logs, as well as an .powerdirector discount strace of the Apache pid. So you’ve checked your .powerdirector discount disk drives and by golly, they are anything but full. Not too surprising as this .powerdirector discount is not referring to disk usage at all, but rather it .powerdirector discount deals with semaphores.

From Wikipedia:

In computer science, a semaphore is a protected variable or abstract data type which constitutes the .powerdirector discount classic method for restricting access to shared resources such as shared memory in a multiprogramming environment. A counting semaphore is .powerdirector discount a counter for a set of available resources, rather than .powerdirector discount a locked/unlocked flag of a single resource. It was invented by Edsger Dijkstra. Semaphores are the classic solution to preventing race conditions in the dining philosophers problem, although they do not prevent resource deadlocks.

So basically a .powerdirector discount semaphore in this context is a unit of resource management. When the .powerdirector discount available number of semaphores has been reached, programs will start to .powerdirector discount fail, throwing the semget error. To confirm whether you have reached the .powerdirector discount current limit, try this one-liner:

ipcs -ls | grep arrays && ipcs -us | grep used

Once that .powerdirector discount has been confirmed, you can track down which user is hogging all those semaphores:

ipcs -s|awk '{print $3}'|sort|uniq -c|sort -n

Often times it will be the same application you’re having problems with, such as the .powerdirector discount Apache user. Now you could reboot the server to clear these out, or .powerdirector discount you can use the ipcrm utility to do it instead. In fact you .powerdirector discount can use the following one-liner to quickly and easily pipe all the .powerdirector discount semaphore ID’s to ipcrm for killing:

ipcs -s|awk '/USERNAME/ {print $2}'|xargs -i ipcrm -s {}

Obviously sub out USERNAME with the .powerdirector discount semaphore bogarting username in question. Hope that helps somebody out there!

Linux Problems Apache, semaphore, semget where to buy CS5 Design Premium

Yeah it’s been awhile since I’ve posted anything. Life has been busy, busy, busy! Anyway, without further delay, on to the tech stuff!

Error: From the command line:

mount error 13 = Permission denied

In /var/log/messages:

Jan 18 00:54:04 hostname kernel: CIFS VFS: cifs_mount failed w/return code = -13

This can occur when trying to mount -t cifs (or mount.cifs) a .powerdirector discount windows share using samba. The basic format for doing this .powerdirector discount is something like:

mount -t cifs //10.0.0.1/data /mnt/data -ouser=username,password=password

The -o flag signals the .powerdirector discount start of the OPTIONS, which here are simply username and .powerdirector discount password. Now, about that error message above. It can mean one of several things:

• The cifs path being mounted does not have a valid share available
• The user specified in the .powerdirector discount mount command does not have access to the windows share
• You need to specify the domain in the username or via the domain= option

For the first point, the easiest way to see what’s available is to use the smbclient command like so:

smbclient -L //10.0.0.1/ -U username

Make sure that .powerdirector discount username is a valid user on the remote windows system. Once you .powerdirector discount hit enter, you will be prompted for the password associated with that .powerdirector discount remote user account. This should output all valid shares, in a .powerdirector discount 3 column format similar to the following:

Domain=[DOMAIN] OS=[Windows Vista (TM) Home Premium 6000] Server=[Windows Vista (TM) Home Premium 6.0]
Sharename      Type      Comment
———      —-      ——-
C$             Disk      Default share
D$             Disk      Default share
data          Disk      Comment (optional, may be blank)

As for .powerdirector discount the second point, regarding a valid user, it is a good idea to .powerdirector discount create a user on the remote windows system just for the .powerdirector discount file shares. No need for this user to have administator access, just make sure they have .powerdirector discount read or read/write permissions to the shares you’re trying to .powerdirector discount mount. The level of access depends on the needs of your .powerdirector discount environment.

Lastly, when .powerdirector discount supplying the login credentials, you may need to specify the domain the .powerdirector discount remote windows user is located in. You can find the domain in the .powerdirector discount smbclient output from above. I’ve had the best luck simply using “user=Domain\username” in the mount command, though you can also use domain=domain,user=username,password=password, etc. Along those lines, it .powerdirector discount is preferable to use the credentials option to specify the location of a .powerdirector discount file with the login information. This helps obfuscate the login credentials so they’re not simply in plain text in fstab.

Speaking of fstab, here is .powerdirector discount how it would look first using the command line and then .powerdirector discount formatting it to work in /etc/fstab:

mount -t cifs //10.0.0.1/data /mnt/data -ocredentials=/path/to/logininfo

//10.0.0.1/data /mnt/data cifs rw,mand,credentials=/path/to/logininfo 0 0

Oh and one last thing, if the remote user has a shebang (!) in their password, this .powerdirector discount may cause problems if entered at the commandline or even in the .powerdirector discount file specified by the credentials flag, due to the way mount is .powerdirector discount apparently quite literal in it’s argument passing. If you .powerdirector discount leave it as is, it fails with the expected “event not found”. Otherwise if you .powerdirector discount try to escape it with a backslash, it seems that the .powerdirector discount backslash gets passed to the remote computer, which obviously fails to .powerdirector discount authenticate. Although there may be some way around this, the time I ran into that .powerdirector discount issue I simply changed the remote password.

I hope this .powerdirector discount helps clear it up for anyone trying to mount their windows shares the .powerdirector discount first time. Feel free to ask for help if you’re running into the above error or any other issue with samba/cifs.

Linux Problems cifs nero on sale, mount, mount error 13, samba

Error: All coldfusion pages (extensions) give 404 Not Found when browsed to.

While you .powerdirector discount might want to rebuild the sandbox and double check the obvious settings in IIS first, this .powerdirector discount is most likely something with the coldfusion mappings directly and can .powerdirector discount be fixed by using the wsconfig runtime utility.  Below is a command line example:

C:\CFusionMX7\runtime\bin\wsconfig.exe -server coldfusion -ws iis -site “mydomain.com” -map .cfm,.cfml,.cfc,.cfswf,.cfr -norestart -v

Here is a break down of what all that means:

• -server (this .powerdirector discount is the JRun servername, if yours is anything other than default, specify it .powerdirector discount here)
• -ws (specify the webservice, in my case it was IIS)
• -site (the domain name or identifier #)
• -map (specify which mappings to .powerdirector discount build, I specified the standards for my environment)
• -norestart (does not restart the web server, no point in this case)
• -v (verbose output, pretty much always a good idea)

I hope that .powerdirector discount helps someone out there in TVLand. It took me awhile of digging before I found it.

Windows Problems .cfm araxis discount code 2011, 404, coldfusion mx7 business contact manager for outlook 2010 down, Windows Problems