|
What is this munin graph showing? I mean, I get that it shows 'apache accesses', but I guess I don't understand what that is. The units seem weird to me. '500 m'? 500 million accesses per second? 500 milliseconds per second? What?
|
#
?
May 5, 2013 18:35
|
|
|
|
| # ? May 19, 2024 17:12 |
|
|
quote:The number of accesses (pages and other items served) globally on the Apache server I don't know anything more than that.
|
|
#
?
May 5, 2013 18:52
|
|
|
I think it 's in milli so 500m accesses per sec is 1 access per 2 seconds.
|
|
#
?
May 5, 2013 18:56
|
|
|
spankmeister posted:I think it 's in milli so 500m accesses per sec is 1 access per 2 seconds. Kind of obtuse, but yeah, that makes sense.
|
|
#
?
May 5, 2013 18:57
|
|
|
Is there a way via stateless autoconfiguration (radvd, specifically) that I can keep a specific MAC address from getting an ipv6 address?
|
|
#
?
May 5, 2013 19:26
|
|
|
nitrogen posted:Is there a way via stateless autoconfiguration (radvd, specifically) that I can keep a specific MAC address from getting an ipv6 address? Even if there's no advertisement visible to the host, it's still going to give itself a link-local IPv6 address. When a host connects to a network, it sends a packet requesting router advertisements. You could ignore or firewall off that packet, preventing the routers from responding. However, all routers will also send unsolicited router advertisements on a periodic basis; these are sent to a multicast address, so there's no easy way to block them from reaching one particular host unless you can setup firewalling or some such on that host.
|
|
#
?
May 5, 2013 19:30
|
|
|
ShoulderDaemon posted:Even if there's no advertisement visible to the host, it's still going to give itself a link-local IPv6 address. Link-local is fine. Let me perhaps explain why i want to do this. My google TV grabs an IPV6 address and has some sort of bug where after a while, it constantly tries to download a shitload of stuff from google's servers, constantly, over ipv6. I'd like to keep it from getting an ipv6 address to see if the problem exists over ipv4 as well. In the stock android config, i'm unaware of any client way to avoid it from getting an ipv6 address. If you know a way to disable ipv6 in a stock (unrooted) googleTV, thats far preferable. Otherwise, I'll probably have to learn how to do traffic shaping or something. Or just get rid of this loving thing. I Read there's a way to do whitelisting on radvd, but i'd rather do it the other way, blacklisting instead. But from what i've read, this is a fools errand anyway, I'm just asking in case anyone else has figured out a way to do this somehow.
|
|
#
?
May 5, 2013 19:51
|
|
|
Ok, since I can't seem to keep my gtv from getting an ipv6 address, I'd like to try and rate limit all its traffic instead. I'm not doing too well here, either. I'm somewhat new to this, so here's what I'm trying to do, and it's not working: code:code:
|
|
#
?
May 5, 2013 20:55
|
|
|
Another approach would be to run your own DNS and filter out AAAA records so it will only use ipv4 when resolving it's DNS names.
|
|
#
?
May 5, 2013 21:11
|
|
|
Or if it's just for a test, disable IPv6 connectivity on that network altogether. I assume you're not putting a GoogleTV on the same network hosting important public-facing servers, so what's the harm if there's no IPv6 service for a few hours/days? Unless you just can't live without the shiny Google logo or the other easter eggs certain sites have when you connect via IPv6.
|
|
#
?
May 5, 2013 22:09
|
|
|
of course it's a production net! its my home network!  Anyway, I got the traffic control working. /sbin/tc filter add dev wan0 parent 1:0 prio 1 protocol ip handle 5 fw flowid 1:5 Apparently, it's always type "ip" even if you are filtering ipv6. It's working now, at any rate.
|
|
#
?
May 6, 2013 01:14
|
|
|
Me and a friend are sharing a VPS running Ubuntu. We have multiple domains through namecheap, using namecheap's dns host records to point everything to our VPS IP and letting nginx handle the rest as far as http is concerned. This meets our purposes fine for http/port 80, but can anyone clue us in on how we'd properly only allow other services/ports to only be accessed via certain domains/subdomains? For example, we have teamspeak running, but as it stands we could access it via either of our domains on port 4272. We'd like to restrict it so the teamspeak server is only reached if the user is attempting to connect via a specific domain. I feel like this should be possible via iptables or hosts but I don't have the correct language for what we're trying to achieve to make googling useful.
|
|
#
?
May 6, 2013 05:22
|
|
|
bloodynose posted:Me and a friend are sharing a VPS running Ubuntu. We have multiple domains through namecheap, using namecheap's dns host records to point everything to our VPS IP and letting nginx handle the rest as far as http is concerned. Doubtful. Once clients resolve via DNS they're all just going to connect to the same IP. You'd need multiple IPs to pull it off, so you can tell service X to only bind to one.
|
|
#
?
May 6, 2013 05:30
|
|
|
bloodynose posted:Me and a friend are sharing a VPS running Ubuntu. We have multiple domains through namecheap, using namecheap's dns host records to point everything to our VPS IP and letting nginx handle the rest as far as http is concerned. Don't think you can do that, as you control the source/destination address etc. in something like iptables. It doesn't inspect the packets to see how the domain was resolved. If you put a hostname in an iptables rules it resolves it when you load it and will have the same result no matter how you got to the IP. Unless you could do something weird with a proxy and pass it along or something, you would want a dedicated IP for that service and move the domain to that (or use a subdomain).
|
|
#
?
May 6, 2013 05:32
|
|
|
The only thing that lets you do vhosts on HTTP is the Host: header, which is a HTTP-specific thing. Unless the applications themselves support some kind of name-based information (almost certainly not, btw) you're not going to be able to switch it out, because nobody connects to hostnames, they always connect to IPs that they resolved.
|
|
#
?
May 6, 2013 05:36
|
|
|
Certain services, and I want to stress that this depends entirely on client implementation so a service with numerous third-party clients may be unreliable at best with this, support the use of DNS SRV records so users only have to enter a domain name but can be told by the SRV record to connect to a non-standard port. It won't allow you to run different services on the same port, but if what you want to run supports it you can at least have different instances on non-standard ports for each user/domain/whatever. TeamSpeak supports SRV records, apparently: http://support.teamspeakusa.com/index.php?/Knowledgebase/Article/View/293/12/does-teamspeak-3-support-dns-srv-records
|
|
#
?
May 6, 2013 13:36
|
|
|
Anybody know of a way to do wildcard cache purges in Squid? I'm aware the official word is "you can't" but I'm sure there's someone else out there that accidentally cached a huge part of their site for 6 months and deleting the entire cache is not an option...
|
|
#
?
May 7, 2013 10:52
|
|
|
I made a typo in /etc/fstab and rebooted and it's throwing me into "disk recovery mode" I need to change 2 loving letters but for some reason the goddamn files are set to read only even as root so I can't change the one thing I need to The one change that would let me boot properly. The one change that it's demanding that I make. Thanks Red Hat. Adult Sword Owner fucked around with this message at 19:45 on May 8, 2013 |
|
#
?
May 8, 2013 19:43
|
|
|
You probably just need to remount as R/W. Most "recovery" type modes I've seen mount disks as RO by default in the interest of protecting any data that may need to be retrieved from the disk.
|
|
#
?
May 8, 2013 19:49
|
|
|
mount -o rw,remount / vi /etc/fstab
|
|
#
?
May 8, 2013 19:56
|
|
|
wolrah posted:You probably just need to remount as R/W. Most "recovery" type modes I've seen mount disks as RO by default in the interest of protecting any data that may need to be retrieved from the disk. Yeah, what he said. If it can't parse fstab then its other option is to mount / and hope for the best. Last thing you want is for anything to write to disk in case it's wrong or corrupt somehow else. Mounting read only is the right thing here. Remount it to read/write then edit fstab.
|
|
#
?
May 8, 2013 19:57
|
|
|
Welp don't I feel like an idiot. Thanks, that worked.
|
|
#
?
May 8, 2013 20:16
|
|
|
Yesterday I made a bridge device and moved the ip config from eth0 to the br0 but forgot to install bridge utils so when I rebooted I locked myself out. 
|
|
#
?
May 8, 2013 20:16
|
|
|
http://sprunge.us/ZPFD Has btrfs finally failed me? 
|
|
#
?
May 8, 2013 22:02
|
|
Associate Christ
|
Kaluza-Klein posted:http://sprunge.us/ZPFD Going to need more context there, stuff like kernel and btrfs-tools version for instance.
|
|
#
?
May 8, 2013 22:19
|
|
|
Well, btrfsck has certainly failed you. I assume you're doing this because your btrfs filesystem no longer mounts? Anyway, this looks like a Fedora 18 live CD. Kernel is 3.6.10-4, btrfs --version gives a spectacularly unhelpful "v0.19". I would compile a new version of btrfsck from git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-progs.git and try that. If all else fails, you'll need to use btrfs-restore to extract your data from what's left of the filesystem.
|
|
#
?
May 8, 2013 22:43
|
|
|
If you have the time and energy you might want to submit a bug report or post on the mailing list so the developers can use your information to make the product better.
|
|
#
?
May 8, 2013 22:45
|
|
|
I think I've googled nearly every possible site on this and I can't seem to come to a conclusion. I'm using ffmpeg to stream to twitch.tv. Everything initially worked out flawlessly then suddenly after a reboot it stopped streaming audio aside from the microphone. Music Players, Games, nothing seems to show up in the stream. I'm not the only one with this problem, ffmpeg seems to be the only choice. VLC apparently is just a fancy GUI scheme for running it through ffmpeg and from what I've read it still has the same problem with audio.code:YouTuber fucked around with this message at 02:11 on May 9, 2013 |
|
#
?
May 9, 2013 02:08
|
|
|
I have a problem that I've posted on ubuntuforums and in the Precise Pangolin thread but I haven't got any replies so I'll try here too. I'm trying to print to pdf through the wordle.net java applet and it's totally unresponsive. Just nothing happens. Is this a bug with java, Chrome/Firefox, CUPS? I can't really find anything about it other than Mac users can't print because Java doesn't have permission to access hardware and there's a fix for that here: http://www.wordle.net/macprint Any help anyone can offer is greatly appreciated.
|
|
#
?
May 9, 2013 10:22
|
|
|
YouTuber posted:I think I've googled nearly every possible site on this and I can't seem to come to a conclusion. I'm using ffmpeg to stream to twitch.tv. Everything initially worked out flawlessly then suddenly after a reboot it stopped streaming audio aside from the microphone. Music Players, Games, nothing seems to show up in the stream. I'm not the only one with this problem, ffmpeg seems to be the only choice. VLC apparently is just a fancy GUI scheme for running it through ffmpeg and from what I've read it still has the same problem with audio. Remember "Monitor of whatever device" is the buildin loopback for system recording. "Whatever device" is the device's microphone. You recording tab should look like this:
|
|
#
?
May 9, 2013 10:50
|
|
|
Has anyone run across a good description of /proc/self/mountstats? The best I've been able to do is look through the source for mountstats.py, but it's still missing a whole pile of fields (specifically, almost all the numbers next to a named RPC call).
|
|
#
?
May 9, 2013 13:11
|
|
|
Does anybody know about running multiple X servers at the same time? I'm pretty new at this, but I was told to run steam (the wine version) in a seperate X server so that alt-tabbing doesn't mess up my games. It pretty much works, but the only problem is, whenever I switch to my main desktop with ctrl alt f1 when playing a game (mostly dota 2), it seems to freeze the game until I switch back, and this generally makes me disconnect from the server I'm in. Is this supposed to happen no matter what, or is there some way I could make it continue running in the background while I switch to my desktop?
|
|
#
?
May 9, 2013 14:08
|
|
|
Illegal Move posted:Does anybody know about running multiple X servers at the same time? I'm pretty new at this, but I was told to run steam (the wine version) in a seperate X server so that alt-tabbing doesn't mess up my games. It pretty much works, but the only problem is, whenever I switch to my main desktop with ctrl alt f1 when playing a game (mostly dota 2), it seems to freeze the game until I switch back, and this generally makes me disconnect from the server I'm in. Is this supposed to happen no matter what, or is there some way I could make it continue running in the background while I switch to my desktop? For gdm you go to gdmsetup->Security->Configure X Server and then you add a second server with the same settings as the first one. For me they end up on ctrl-alt-f7 and ctrl-alt-f8. I usually use those to try out fancy window manager stuff, so I just log in the second server with a different wm than in the first one. If you want to go for maximal separation you could also create a second user to log in on the second server.
|
|
#
?
May 9, 2013 15:00
|
|
|
Illegal Move posted:it seems to freeze the game until I switch back, and this generally makes me disconnect from the server I'm in. Is this supposed to happen no matter what, or is there some way I could make it continue running in the background while I switch to my desktop? Unfortunately, this isn't possible without fixing the X server. A lock is taken when VT switching that pauses all GLX clients. Running your games in another X server isn't really something you should be doing.
|
|
#
?
May 9, 2013 15:09
|
|
|
Misogynist posted:Has anyone run across a good description of /proc/self/mountstats? The best I've been able to do is look through the source for mountstats.py, but it's still missing a whole pile of fields (specifically, almost all the numbers next to a named RPC call). I took a look at it on my CentOS 6.4 box and it looks like /etc/mtab but I don't see any numbers in /proc/self/mountstats . Maybe something like nfsstat is what you're looking for?
|
|
#
?
May 9, 2013 15:31
|
|
|
Any SELinux people out there? What's the difference between doing a "semanage fcontext" and then a "restorecon" vs doing just a "chcon"? If I understand it right doing semanage then restorecon somehow permanently stores the semanage change ("/foo's default is now public_content_t") and then the restorecon restores it to it's default (now public_content_t). chcon just straight up changes it and if you were to do a restorecon after the chcon it would restore it to whatever the current default is (say public_content_rw_t or whatever). Is that correct? The "proper" way is to semanage then restorecon, but chcon would essentially give the same end result minus changing that directory's default context.
|
|
#
?
May 9, 2013 15:40
|
|
|
hackedaccount posted:Any SELinux people out there? What's the difference between doing a "semanage fcontext" and then a "restorecon" vs doing just a "chcon"? If I understand it right doing semanage then restorecon somehow permanently stores the semanage change ("/foo's default is now public_content_t") and then the restorecon restores it to it's default (now public_content_t). chcon just straight up changes it and if you were to do a restorecon after the chcon it would restore it to whatever the current default is (say public_content_rw_t or whatever). You can semanage then chcon as well. The real difference is that you'll have a bad time if something relabels the filesystem after chconning a bunch of files.
|
|
#
?
May 9, 2013 16:46
|
|
|
Lysidas posted:Well, btrfsck has certainly failed you. I assume you're doing this because your btrfs filesystem no longer mounts? Yeah, mounting segfaults, so I was using an F18 live cd. koji thankfully had a btrfs-progs rpm built from git on May 1, but it does no better. I would love to dig deeper into why this happened, but this is my work laptop and I need it to be back on track asap. I don't even have a spare disk large enough to dump a btrfs restore or dd on, anyway edit: I just remembered my wife's macbook has a 'giant' 250GB hard disk that is mostly empty. I have mounted it as a cifs share and btrfs restore is seemingly dumping files galore into it. All is not lost, it seems! I really wish I knew if this whole thing was a freak accident or if it relates at all to the new SSD I am using... other people fucked around with this message at 18:18 on May 9, 2013 |
|
#
?
May 9, 2013 17:59
|
|
|
hackedaccount posted:minus changing that directory's default context
|
|
#
?
May 9, 2013 18:12
|
|
|
|
| # ? May 19, 2024 17:12 |
|
|
evol262 posted:You can semanage then chcon as well. The real difference is that you'll have a bad time if something relabels the filesystem after chconning a bunch of files. Perfect, thanks.
|
|
#
?
May 9, 2013 18:23
|
|