Server :: Mount Application Filesystems Only After OS Is Loaded?
Apr 13, 2011
I have servers where the disk is partitioned into OS and application partitions. The application is rather sloppy with its file handles, and frequently when there is a system crash, its file systems get corrupted. This will cause fsck to halt the boot, requiring me to get on a remote console, enter the root password, and fsck the file systems.
Is there a way to have just the OS partitions get checked and mounted, then check the application systems, only after the OS is loaded, so I can ssh into the system, instead of having to use a bandwidth-hogging remote console?
The app partitions are already at fsck level 2 in /etc/fstab, but this doesn't prevent OS loading from halting. If I set the fsck level to 0, I don't want the application to start if the partitions are unavailable. Should I just leave the fsck level at 0? Or should have the partitions marked "noauto", then have a startup script run fsck and mount the app partitions?
i have purchased sun server for my visualization project. Request you to help me finding the io for disk .I have put storage(disks) on different location (File Server) and on server (Application) i have configured 4 virtual machines.
How would i monitor the io for file server from the Application server where i have configured 4 virtual machine. one way to mount the file server share on application server and execute
dd if=/dev/zero of=/share/test.out bs=1024 count=1024 to check the read and write Is there any other way of doing this.
One of the good points of linux is that is easy to customize the partitioning scheme of the disk and put each directory (/home, /var, etc) in diferent partitions and/or diferent disk. Then we can use diferen file system/configurations for each of them for make them better. xamples:
noatime is a mount option to not write access time on the files. data=writeback is an option to layz write metadata on new files. ext3/4 has journaling that make the partition more secure in case of a crash. bigger blocks make the partition waste more space, but make it faster to read and may become more fragmented. (not sure) Then: What are the best filesystem/configurations for each directory? Note: given the answer of Patches, will only discuss /, /home and /var only.
/var -> It's modified constantly, it write logs, cache, temporal, etc. /home -> stores important files. /-> stores everything else (/etc and /usr should be here)
I'm setting up an Ubuntu server to replace my aged Pentium IV Slackware box. It's a Dell Inspiron 560 with modest core-2 duo processor, 8 gigs of ram, and a pair of good sized hard disks. I came upon a good deal on a couple of 40gig Intel SSDs. I'd like to use one in the server. I'd like to use the SSD for the relatively invariant stuff, because they write slow, and are life-limited in the # of writes. So: /bin /usr/bin /boot /etc /lib /usr/lib /usr/local/lib /mnt /opt
The best way IMHO to achieve this would be to make the SSD the root, and mount hard drive partitions/filesystems to it to places such as: /var /media (Here you read and write giant files. Hard disks do this just fine. One will work especially fine if one particular hard drive is dedicated to this.) /root /home /tmp
A quick "df" yields a list of filesystems. There are four that are not tied to any device! /dev /dev/shm /var/run /var/lock
(df also discloses that the root filesystem is presently standing at 502megs. Guess it'll fit in a 40-gig SSD). These deviceless filesystems worry me. Are they created magically on boot? What's required to make the system magically create them on boot? If I copy the filesystem over to the SSD and redo the grub config, will it Just Work? Web searches reveal subtleties WRT mount points.
I can't get Redhat (Fedora) to mount my usb Flashdrive. My windows O/S has no problem with my usb flashdrive. I can mount my CD Drive and my Floppy drive. My mouse is pluged into one of my usb slots and it works. What am I doing wrong.
We've installed RHEL 4 (Linux Redhat) on a virtuel machine VMware. This virtuell machine has 2 CPUs (CPU0 and CPU1). We also installed the JBoss on it. But we notice, that in run time the system load of one CPU is quite often 100%. I am not sure if this is caused by the application running on Jboss or there is error by configuration on Linux.
what is happening on my Centos virtual server (ps I'm having a great time re-learning Linux after a large gap and usually try and find out myself but this has me confused) First thing I did was install Tomcat and try and change the port to 80 using iptables -nat and it complained a kernel module was missing. Instead setup Apache to redirect requests so that's now working I then tried to install CSF firewall and the validation perl script returned fatal errors so I assumed OK my virtual server Kernel is restricted and used denyhosts instead which is working great
Here's the confusion.. I later retried the csftest script and it now works including all the iptables and nat stuff that fataled before. I've not rebooted or changed anything except run denyhosts and the postgres daemon + ended a few daemons that I didn't need (eg pickup,tlsmgr,courier,postfix,qmgr,saslauthd) Running lsmod returns nothing and /proc/modules is zero bytes so I'm not able to tell what is loaded? I'm tempted to try and get Tomcat redirecting via iptables and see what happens but if it works I don't like that something seems to have changed.. before I ran it direct as root now I'm SU into root from another login
I have /var/log/audit and /var/log/audit.log owned by root and 600 permissions. I've also removed and made an empty /var/log/audit directory when that did not we work either. I can start the service after boot up, but it is not coming up automatically even when configured by chkconfig. I also get this after I attempt a restart...
Stopping auditd: [ OK ] Error deleting rule (Operation not permitted) Starting auditd: [ OK ] The audit system is in immutable mode, no rules loaded
A tail of my /var/log/messages shows this... Nov 23 16:45:18 hostname kernel: type=1302 audit(1290548718.524:73): item=1 name="/var/run/auditd.pid" inode=131143 dev=fd:01 mode=0100644 ouid=0 ogid=0 rdev=00:00 obj=user_u:object_r:var_run_t:s0 Nov 23 16:45:18 hostname kernel: type=1300 audit(1290548718.618:74): arch=c000003e syscall=87 success=no exit=-2 a0=7fff730b2f85 a1=7fff730b2f85 a2=2 a3=0 items=1 ppid=6243 pid=6248 auid=1111 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=0 fsgid=0 tty=pts1 ses=1 comm="rm" exe="/bin/rm" subj=user_u:system_r:unconfined_t:s0 key="delete" Nov 23 16:45:18 hostname kernel: type=1307 audit(1290548718.618:74): cwd="/" Nov 23 16:45:18 hostname kernel: type=1302 audit(1290548718.618:74): item=0 name="/var/run/auditd.pid" inode=131073 dev=fd:01 mode=040755 ouid=0 ogid=0 rdev=00:00 obj=system_u:object_r:var_run_t:s0 Nov 23 16:45:18 hostname kernel: type=1300 audit(1290548718.620:75): arch=c000003e syscall=87 success=yes exit=0 a0=7fff9b776f81 a1=7fff9b776f81 a2=2 a3=0 items=2 ppid=6243 pid=6249 auid=1111 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=0 fsgid=0 tty=pts1 ses=1 comm="rm" exe="/bin/rm" subj=user_u:system_r:unconfined_t:s0 key="delete" Nov 23 16:45:18 hostname kernel: type=1307 audit(1290548718.620:75): cwd="/" Nov 23 16:45:18 hostname auditd[6260]: Started dispatcher: /sbin/audispd pid: 6262 Nov 23 16:45:18 hostname audispd: af_unix plugin initialized Nov 23 16:45:18 hostname audispd: audispd initialized with q_depth=80 and 1 active plugins Nov 23 16:45:18 hostname auditd[6260]: Init complete, auditd 1.7.17 listening for events (startup state enable)
My questions is...just like on the subject, what is the best applications to mount ISO image (which is not an OS image) in linux. perhaps the most user friendly.
Does anybody have experience with linux tuning. I`m realy interesting about sysctl.conf tuning settings for batch(3d rendering, phisycs simulations, etc.) servers. Does anybody has an experience with linux tuning - I mean memory and CPU settings for heavy loaded systems. What kind of settings You have in Your clusters ? I`m working with Red Hat Enterprise 5 x86_64.
I have just been told that httpd was not running on my Centos virtual server. It seems that my hosting company rebooted my server 2 days ago and that httpd restart failed due to:
Starting httpd: httpd: Syntax error on line 213 of /etc/httpd/conf/httpd.conf: Syntax error on line 2 of /etc/httpd/conf.d/proxy_ajp.conf: Cannot load /etc/httpd/modules/mod_proxy_ajp.so into server: /etc/httpd/modules/mod_proxy_ajp.so: undefined symbol: proxy_module
I had worked around and forgotten about this some time ago. The problem arose because the most recent httpd update installed proxy_ajp.conf into /etc/httpd/conf.d which was then loaded on restart The trouble is I do not use mod_proxy (I think just to try and get a lean and mean web server) and proxy_ajp.conf does not check if mod_proxy is loaded before it tries to LoadModule proxy_ajp_module modules/mod_proxy_ajp.so
Hence httpd fails to restart. RedHats's bug tracker lists this as an old issue (which was never resolved) [URL] My workaround was to rename proxy_ajp.conf to proxy_ajp.conf.orig so that httpd would start up correctly. However any updates simply reinstall proxy_ajp.conf and so IMO the problem needs to be resolved at source. how to ensure that httpd updates do not bite me in the future? Do I really have to run mod_proxy whether or not I need it?
is there a way to allow a program mounting a drive without requiring it to ask for sudo password (apart from running it with sudo)? To be more specific, I'm annoyed by the sudo password request by TrueCrypt whenever it needs to mount a volume.
I've thought about creating another user, allowing it to mount volumes and then running TrueCrypt as this user at boot. I don't know whether GNU/Linux allows for such policies... maybe I should look into SELinux?
EDIT: For the issue at hand (encrypted USB stick both on Windows and Linux), I'm investigating FreeOTFE. I'm still curious about the privileges issue, anyway.
I'd like to make regular backups of my flash drives, but I don't plug them in regularly. Because of this, I can't just use a scheduled backup application to back them up. Does anyone know of an incremental backup application that will automatically run backups on media as soon as it's mounted?
I've tried to get an opensuse box I have to share a directory via NFS. I've failed each time, but I thought that the third time, I'd enlist some help from the forums, if I could. how do I know that the nfs server and not the client is the problem? Short answer is: I don't! That's why nfs (and many netwrk problems) are laborious, you're troubleshooting needs to take place at both source and desitination. Next question, what do I have set up so far? Well, I did download the nfs server kernel stuff (two months back) and /etc/init.d/nfsserver start seems to get set up OK. No errors and the daemons nfsd, idmapd, mountd area all running. So, I *think* that part is OK. I have the share set up properly in /etc/exportfs and have "exportfs -r" it.
OK, now onto the trickier stuff: the client and iptables. On the client pinging to the nfserver box is perfect, and I have rpcbind running. the reported error is "mount.nfs: mount system call failed" though from experience nfs errors don't mean a whole lot.However, I will go off and check now and see if I need a mountd running on client-side too.Then there's iptables .... ouch, that could be a long and painful trek. I don't see any specific ports being blocked, and it's the iptables that the default v11.2 opensuse came with. I did turn them off and the problem was the same, so whether wishfl thinking or not, I'm hoping it's not an iptables issue.
i have installed nfs server on my redhat machine.when i want to mount shared data from client(suse)machine the following error occur."mount.nfs: mount to NFS server '10.3.31.146:/home/usbtest' failed: System Error: No route to host"
I have servers installed with RHEL 4 2.6.9-89.0.9 ELsmp. I tried using uuid and label in /etc/fstab to automount usb drives to mountpoints that I specify after reboot. Unfortunately, it just does not work in all my RHEL4 servers. After every reboot, /etc/fstab will be automatically modified and all configurations related to my USB drives will be changed. Irregardless of whether i use UUID or LABEL in my /etc/fstab.However, it works on RHEL5. But, upgrading is not an option in my environment. I have been googling around looking for alternatives but everything seems to point back to using UUID or LABEL in /etc/fstab. Anyone has tried something that works? Please help me, thank you.
I have a query on mod_rewite/apache/.htaccessIt is like this:Suppose a company have a website www.abc.com,when a user types in this URL in his browser he should get a login and passwd prompt say he enters username as user1 and some passwd, upon successful authentication he should get redirected to another url similary for user2 and user3...mod_rewite is loaded htpsd.conf has "rewrite engine on" directive enabled.
I have a web application(java 5, mysql server) installed on Tomcat on Red Hat 5.Part of the functionality is to allow the upload of files to a windows share. I have mounted the share using theAny file I create here from the terminal gets sions."-rwxrwxrwx"The problem is that when the application server creates the file it gets created with the following."-rw-r--r--"How can I get files created from the server to be "-rwxrwxrwx"
Can anyone tell me in performance tunning of apache-tomcat and jboss application server?when I deploy some application in apache-tomcat the performance is say 100 users per second and incase of jboss it is even worst (35-40 users per second)i want to improve my performance 1000 uesrs per second..
I am trying to make an application server that would contain simple ones like edit, openoffice, gimp, so on and so forth. I know how to access these using ssh but what I want is that the application runs on that server as it already does, but to open local files and save them locally.
We are trying to define an appliance for an application server so I would like to know which should be the best file system type for this kind of use, basically our web applications uses libraries of 50 KB and our web apps.creates temp and logs files not bigger than 3 MB.
We recently shifted our applcation and database to LINUX from UNIX.so now application server(weblogic 8.1) is on linux and oracle database 9i is on linux.previously there is a process in application server(which access the Database dor the data) which took only 1 hour to run on UNIX after shifting to LINUX the same process is taking 4 to 5 hours to run.
but when individually ran the queries on the database it is quck than UNIX.our ADMIN tried changing the kernel parameters for the database server, but it is still the same.
I want to know what file system there are for "/dev" directory??apparently different types was developed for managing devices on Linux.because I am a little confused about all file system on this directory.Another question is,which is file system sufficient for managing devices on Embedded Linux ?of course on our embedded system we do not have many pluggable device,so this directory can be static
I'm now using Ubuntu 10.04 with ext4 and for the second time in my life I experienced data loss (not for real: I got backups) and I'm assuming problems with the recent ext4 fs.
I want to restore all of my configurations (/etc and the like), data and home on reiserFS: is this possible? What to do in order to accomplish that?
I am tired of watching fsck check my filesystem when my eeepc 901 shuts down abruptly due to a crash. I know that with a journaling filesystem, I won't have to wait for a check. However, I am well aware of the poor I/O performance of the SSD, so I can imagine using a journaling filesystem being even more frustrating, since there will be constant writes to the journal? I will buy a new laptop without such a crummy ssd someday but, is there anything I can do now, on the software side of things?
