|
odiv posted:I have the following line in the rc.local file of my linux router: There are ways to filter what gets reported: you can choose at least what port you don't want to see. I believe that it's under the admin section.
|
# ¿ Sep 14, 2008 16:45 |
|
|
# ¿ May 10, 2024 10:10 |
|
Hi everyone, I'm having trouble mounting a cifs share on a debian box that we have at work. It houses our wiki and I want to offload the backups I make. At this time I have 2 other systems centos boxes that are also connected to the share and can see files. When I try to connect with the debian system I get: code:
code:
Thanks!
|
# ¿ Sep 18, 2008 19:58 |
|
Hey everyone, I have a local yum repo for our servers at work, and I use rsync to grab updates from the university of Oregon. However, much of the time is spent grabbing packages that I don't need like Cluster_Administration in 462356235 different languages. is there a way that I could use regex or something similar to ignore certain sets of packages. here is the 1 liner code:
|
# ¿ Oct 1, 2008 15:35 |
|
Gumball Dad posted:Oh boy! I got MySQL installed successfully. Turns out I didn't have to resolve every dependency, I resolved one and everything magically worked. I'm not one to question providence. Now I get to troubleshoot MySQL. I've got the daemon running, but when I try to ./mysqltest nothing happens. Am I even going about this right? How do I actually use MySQL? Like run the program? I would suggest going to the mysql website and following the tutorial. http://dev.mysql.com/doc/refman/5.0/en/tutorial.html This will take you through some of the basic commands and administrative duties that running mysql needs. Also here are some mysql cheat sheets which you might want to reference. http://www.addedbytes.com/cheat-sheets/mysql-cheat-sheet/
|
# ¿ Nov 7, 2008 02:08 |
|
Sock on a Fish posted:host, ping, dig, etc. all report that they're using the servers specified in resolv.conf, and nsswitch.conf hasn't been modified. nscd is installed and running, but that just provides a local cache, right? That would explain hosts not resolving correctly, but not the presence of DNS queries going to servers not specified in resolv.conf. You might want to try and put the new server in /etc/hosts/. I've seen some odd issues resolved (lol) by this. Ymmv though.
|
# ¿ Nov 26, 2008 20:30 |
|
Hello everyone: I got an interesting question regarding permissions and vsftpd. I have a ftp folder that will have 2 groups of people accessing it. Group 1 needs full read/write permissions on that folder. Group 2 needs only read access to that folder. While I have Group 1 being able to access download and upload files. I'm unsure as how I would force Group 2 to only have read permissions. Any permissions experts have any ideas?
|
# ¿ Feb 11, 2009 15:06 |
|
I've started on the journey to automate many of the server builds at work using kickstart. However before I do this I want to create some custom RPM files so I can automatically build needed packages. I've created an rpm file using the rpmbuild command, however I'm a bit confused as to where I can add my custom configuration files into the spec file. I know there is a %config macro but I'm unsure as to where I should put the configuration files for building. Here is my kickstart file code:
|
# ¿ Apr 20, 2009 20:29 |
|
juggalol posted:Just for curiosity's sake, I ran 'hdparm -i' on an (internal) SATA disk and was able to get the serial number. Running the same command against an external USB disk gave me an error message, "HDIO_GET_IDENTITY failed: Invalid argument". You might want to try the -I flag, it seems that the -i flag is for older devices. hdparm man page posted:-i Display the identification info that was obtained from the drive
|
# ¿ May 14, 2009 14:12 |
|
I have a centos 5 server which needs to be upgraded past the 2.2.3 version of httpd that ships in the centos repos (due to bug fixes that aren't being back ported for what ever reason) I'd like to mirror the configuration flags that centos uses when it installs the httpd package, but I'm having trouble finding them. Is anyone aware of a location where those flags might be saved?
|
# ¿ Jun 2, 2009 16:18 |
|
JHVH-1 posted:You could probably get the configs by downloading the source rpm of the package you have installed. There is a spec file that it includes so if you install the src RPM it the source and spec file shows up somewhere in /usr/src/redhat/ Great idea, and this worked perfectly I used yumdownloader --source (package) to grab the source package then installed it. thanks a lot!
|
# ¿ Jun 2, 2009 17:24 |
|
Anunnaki posted:How do you change the time format to a 24-hour clock? I changed it for the clock that's in the upper-right corner, but things like my Thunderbird calendar, and IM timestamps still show up with the 12-hour AM/PM stuff. I can't find a setting for this anywhere, and Google isn't much help. I know in KDE (this is centos btw) you can just right click on the clock and select date&time format.
|
# ¿ Jul 3, 2009 14:08 |
|
Sharrow posted:[debian/ubuntu] I've been using cfengine as well as mercural for config change management give that a look.
|
# ¿ Jul 20, 2009 17:23 |
|
dreggory posted:Not sure if this warrants its own shiny thread so I thought I'd play it safe and ask here first. You could try some sort of VPN solution this would allow more private access than you would have using just SMB/CIFS over the internet.
|
# ¿ Aug 5, 2009 21:44 |
|
n0manarmy posted:If I've got no choice but to use a distribution that supports RPM's is CentOS still the reigning king for free? I set up OTRS on a centos box for our company hit me on a pm if you have any questions.
|
# ¿ Aug 14, 2009 18:09 |
|
Jerk McJerkface posted:
|
# ¿ Sep 2, 2009 19:52 |
|
FunOne posted:I'm having a weird issue with Firefox on CentOS 5.1/2 (whatever, fully updated). I occasionally get huge delays in resolving DNS names. I can open a terminal and nslookup just fine during these periods, and I don't have similar issues on any other machine (Windows) on the network. Have you disabled ipv6 on the interface?
|
# ¿ Oct 21, 2009 23:49 |
|
Severed posted:Thank you, working like a charm now. Can you ping the box? Are you working through a router or anything like that? Is iptables on blocking the port, or if it's on are you enabling access? On the subject of VNC, I've found that http://www.nomachine.com/ works really well for these types of things, and it's a bit more "fire and forget" if you would. If you continue to have trouble with setting up VNC, give this a try. Twlight fucked around with this message at 22:19 on Nov 7, 2009 |
# ¿ Nov 7, 2009 22:14 |
|
Mustach posted:I'm trying to set up a cron task that sends out an e-mail to a list of people. The e-mail script looks something like this: I've had issues with running scripts from cron tabs that exhibit the same issues. Usually it is because commands in your program aren't explicitly defined. Try making everything in the script explicit (like /bin/mail or where ever it's located I'm not sure off the top of my head).
|
# ¿ Dec 5, 2009 18:36 |
|
Misogynist posted:It depends on your environment. You're very unlikely to get support from the vendor if you have a customized kernel, so they're uncommon in most of the world, but in a lot of high-frequency trading shops you will see real-time kernels in the datacenter. This is true, I work with a few different ISV's and they all have their own RT kernel that we put together for them. Sometimes it doesn't play nice with different hardware, Dell R710 I'm looking at you!
|
# ¿ Mar 17, 2010 17:27 |
|
I have a client with a very old system and he is running some legacy apps that have been written in C++. However he is asking that we install ipmitool so as to allow him access to his ILOm. Installing the ipmitool results in the following error: configure:6679: checking whether we are using the GNU C++ compiler configure:6703: g++ -c conftest.cc >&5 g++: error trying to exec 'cc1plus': execvp: No such file or directory configure:6709: $? = 1 After some googling, I made sure that both g++ and gcc were the same version but it seems that I just need to update g++ and or gcc, which in it's self is no big deal. However this server is running some legacy applications written in C++ which may or may not be using libraries on the system my questions are these: If I upgrade G++ / GCC would that over write the existing libraries? is there a way to tell what libraries are in use? Thanks.
|
# ¿ Mar 30, 2010 14:28 |
|
Rastor posted:You could probably even get away with compiling it staticly on a non-legacy system and then copying over the binary. Good thinking I'm going to give this a try. Standish thanks for the tip on ldd.
|
# ¿ Mar 30, 2010 17:36 |
|
I have an odd question, I do a lot of installs on servers that have multiple nic's (from 8-18 physical ports) across both Intel and broadcom cards (broadcom usually being on boards). However I've noticed that once the install is finished the OS (SUSE, Red Hat, Centos all have showed this behavior) the OS will "mis align" the nics, making the Intel cards come up as eth0,eth1 etc. However, due to internal SOP, I would like to make the broadcoms come up eth0,eth1 etc etc. I know that the Intel driver is loaded before the broadcom driver during the install, is there any way using kickstart or some sort of install flag that can force the broadcom drivers to come up first, or at least before the Intel drivers.
|
# ¿ May 17, 2010 17:19 |
|
ShoulderDaemon posted:I wouldn't bother with trying to get drivers to probe in a particular order, that's fragile and prone to failure down the line. Just use udev's functionality to give particular cards (identified by bus ID or mac address) the correct names, or do something with ifrename or ip link set name if for some reason you can't use udev. Right, usually we switch the mac addresses around to force them to come up in the correct order. What I was thinking was using anaconda's blacklist feature to stop the install of the igb drivers then installing them later from the disk after the broadcoms have been setup. This way will reduce our time to get things set up, and will make sure that things stay in the right order.
|
# ¿ May 17, 2010 18:08 |
|
ShoulderDaemon posted:I can't imagine how futzing with driver load order could possibly be better than just renaming the interfaces once they're all up. Linux doesn't make any promises about how devices are probed and what order they come up in; that's the entire reason we need userspace tools like udev to dictate naming policies. If you try to enforce the ordering by altering when various modules are loaded, then you are risking it all falling apart whenever you upgrade the kernel or have to swap out a motherboard or if there's a glitch and the kernel decides it needs to reenumerate the PCI bus or one of the cards dies and now all the interfaces are wrong instead of just one being missing or any number of other situations that are very difficult to plan for in advance. If you assign names based on mac addresses, this just doesn't happen, and the device that you want to be eth2 will always enumerate as eth2, even if eth0 and eth1 aren't ready yet. I can see where you're coming from to let the kernel just do it's thing, I was wondering due to the fact that this happens multiple times and is a pain in the rear end with the amount of NICS in play as well as the order. A question however: I mentioned the blacklist anaconda blacklist flag to flag the Intel driver during install then installing it later. What would be the negative to doing this, If I'm going to be installing the driver anyhow. Not that I'm trying to argue, I'm curious as to what this might break.
|
# ¿ May 17, 2010 18:31 |
|
ShoulderDaemon posted:Edit: I mean, honestly, the work required to do the correct thing here is literally no more than adding a udev rules file with lines of the format
|
# ¿ May 17, 2010 18:34 |
|
ShoulderDaemon posted:In the Intel case in particular, in addition to being subject to the whims of newer kernels deciding to probe things differently or hardware faults or what-have-you, you also get to deal with the fact that some Intel cards are at least partially supported by multiple drivers, so if you blacklist the driver it's supposed to use, you may wind up with it enumerating anyway in 10mbps mode or something with an alternate driver. There are a few other classes of device this happens to; SATA cards and a number of wireless cards, especially. This is one of the reasons you should really only blacklist a module if you are having a problem with that module, not to try to reorder the way the kernel does enumeration; the tools assume you want your hardware to enumerate as quickly as possible, and will fight you. Interesting thanks Shoulder.
|
# ¿ May 17, 2010 18:39 |
|
Did some googling and came up with this. I guess drives over 2TB need to use GPT as their labling schema, in this case centos doesn't like disks that large take a peek while it doesn't mention red hat out and out this might help http://idolinux.blogspot.com/2008/10/perc-raid-and-efi-gpt-wierdness.html
|
# ¿ Jul 3, 2010 05:49 |
|
This is an odd one i've come across. I've been working with a client to tune their nic's we're using the IGB driver in SUSE 10.2 x64_86 here is our version: ?!?!?:/etc # ethtool -i eth7 driver: igb version: 2.1.1 firmware-version: 1.5-1 bus-info: 0000:08:00.1 ?!?!?:/etc # I'm trying to change the coalesce paramaters, of note, the tx-usecs paramater: !?!?!!?:/etc # ethtool -c eth7 Coalesce parameters for eth7: Adaptive RX: off TX: off stats-block-usecs: 0 sample-interval: 0 pkt-rate-low: 0 pkt-rate-high: 0 rx-usecs: 13 rx-frames: 0 rx-usecs-irq: 0 rx-frames-irq: 0 tx-usecs: 0 tx-frames: 0 tx-usecs-irq: 0 tx-frames-irq: 0 rx-usecs-low: 0 rx-frame-low: 0 tx-usecs-low: 0 tx-frame-low: 0 rx-usecs-high: 0 rx-frame-high: 0 tx-usecs-high: 0 tx-frame-high: 0 I've changed the rx-usecs number by adjusting the InterruptThrottleRate inside of modprobe.conf doing this i'm able to change the rx-usecs. You can also change this value using ethtool. however when trying to change the tx-usecs values using ethtool: ?!!?!?:/etc # ethtool -C eth7 tx-usecs 13 Cannot set device ring parameters: Invalid argument I was wondering if anyone has changed these values using the IGB driver and was successful.
|
# ¿ Jul 19, 2010 21:58 |
|
Kilson posted:This doesn't actually work (hence my question in the first place), you still get the gateway address only. Can you show a route table? as well as the commands your using to route?
|
# ¿ Jul 30, 2010 18:15 |
|
ok just so I know what's going on here: you've got multiple VLAN's that you want to push through the same nic right? if this is the case you can do Linux vlans then route through the vlans, so using your route command above you can vlan (subinterface) then do the dev eth0.1 or what ever. If I'm off base in this case what is to be done?
|
# ¿ Jul 30, 2010 20:10 |
|
dr go hog wild posted:I have a weird problem. Kinda moot since there's a workaround, but: Try seeing if you can do a tcpdump on the wlan0 interface to see anything more. Might give you a bit more information, though DHCP issues can be extra annoying, Do you have access to the router to see if you're showing up under known hosts?
|
# ¿ Dec 21, 2010 16:53 |
|
HatfulOfHollow posted:I had this graphic hanging in my cube for a long time as a reference. It's pretty solid and organized well Thanks for this.
|
# ¿ May 2, 2011 21:56 |
|
you can put your routes in /etc/init.d/ then make a symlink to rc3.d and or rc5.d so they'll load on which ever runlevel you use.
Twlight fucked around with this message at 20:23 on May 31, 2011 |
# ¿ May 31, 2011 20:19 |
|
Misogynist posted:This is technically true, but also more or less what rc.local was designed for. And here I was doing more work than I needed to. Thanks Misogynist.
|
# ¿ Jun 1, 2011 17:35 |
|
JHVH-1 posted:I haven't tried it, but this is the kinda crazy thing that might be cool http://www.fudgie.org/ This is loving raw, though I'm sure after about 10 min of a busy webserver you'd get a headache.
|
# ¿ Sep 23, 2011 16:15 |
|
ClosedBSD posted:All the cool cats are on IRC, but they are pretty much dismissive pricks. Your best bet for a Linux "community" is, sadly, probably the mailing lists, or maybe the Arch Forums. Whats the fresh Linux IRC community these days?
|
# ¿ Oct 12, 2011 19:10 |
|
spankmeister posted:Sure, if you do Also to this you can add the -y flag to yum, this will just automate it to say yes to any prompts ( for pgp keys or are you sure you want to install prompts )
|
# ¿ Jul 13, 2012 18:09 |
|
River posted:^^ Same result. I might just bite the bullet, backup everything and reinstall if I can't find a way to fix my scripts. I'd use du or something but it's a pain. Just for a lark, try doing this: tune2fs -l </this/partition/> look for "Reserved Block Count" Many file systems will reserve space so that a normal user cannot fill the root partition to 100%, causing some issues. I just had to deal with finding some extra space that was "hidden" and found it with this. Not that reserving some emergency space isn't a good idea. But it might help solve the mystery.
|
# ¿ Feb 26, 2013 14:45 |
|
Dameius posted:Lets assume that I am rusty as hell on coding, but can basically follow along on simple stuff (because it is true). Lets also assume that I need to as quickly as absolutely possible make Linux second nature (also because it is true). I'll be using Fedora 16 mostly. Besides putting Linux on a VM or a throwaway partition and just start messing around in it what are the best ways to very quickly crash course my way through learning Linux backend? Any guides or pdfs or even books that can supplement hands on learning with free being the optimal price but I can shell out some cash if it is particularly worth it. I think the best way is to make things that are relevant to your life or something you'd like to do. Create an emulator, host some media, have a mail server, host a webpage, install cacti(graphing for system statistics) things like that. I always find that it's far harder to learn something by force when you don't have any sort of personal investment. Just going through some webpages about how to do things will only teach you that, how to do it. Not why it's done or why it's done for x y and z reasons.
|
# ¿ Mar 15, 2013 12:52 |
|
|
# ¿ May 10, 2024 10:10 |
|
Martytoof posted:Uhhh, I'm having sort of a brain fart moment. I gave this a try: :~$ PLATFORM=$(uname -s) :~$ echo ${PLATFORM,,} linux And it spits it out alright, I never knew about the ,,'s either that's neat.
|
# ¿ Sep 13, 2016 15:12 |