=== hendry1 is now known as hendry | ||
Hamidreza | Hi, I would appreciate if someone help me on this : https://pastebin.ubuntu.com/p/VwVWMG39cf/ | 10:18 |
---|---|---|
RoyK | Hamidreza: check the output of ss - netstat is deprecated - anyway, I guess this just means it's listening to both protocols, which sometimes show they are merged, somehow. You really cannot connect to ipv4 with ipv6. That'll be like talking greenlandic to an englishman - doesn't work too well… | 10:50 |
Hamidreza | RoyK, so why it is binding like this? I'm giving the program the exact name | 10:52 |
RoyK | Hamidreza: did you check output from ss? | 12:19 |
=== vlm_ is now known as vlm | ||
qman__ | Hamidreza: linux supports a feature where programs can bind on both ipv4 and ipv6 on the same port, and it shows up as an ipv6 bind, but the kernel still accepts ipv4 connections | 13:13 |
qman__ | the weirder thing in your post is that you have two binds on the same port, which should be impossible | 13:13 |
qman__ | also, ss is a drop-in replacement for netstat, accepts the same arguments, the output is slightly different though | 13:13 |
sigv | qman__: SO_REUSEPORT allows re-using a port since Linux 3.9: https://lwn.net/Articles/542629/ | 13:14 |
qman__ | strange, but ok | 13:18 |
qman__ | in any case, Hamidreza, the app shown there is binding on *:8080, using the aforementioned feature, so the bind shows up as :::8080 and is dual stack | 13:20 |
qman__ | 8080 is typically used as an HTTP alternative port, so I wouldn't expect a database to use it, but it's not impossible; if it is, though, it's not binding on a specific IP, it's binding on any IP | 13:21 |
cpaelzer | ijohnson: glad I could help with that apparmor issue | 13:51 |
cpaelzer | ijohnson: is there any further follow on needed from me? | 13:52 |
ijohnson | cpaelzer: hello, no I think I'm good I know what the issue is and how to work around it now, so everything is good again for me :-) | 13:54 |
ijohnson | well at least when it comes to using swtpm + qemu, the rest of the world maybe not quite so positive, but that's an orthogonal concern :-) | 13:54 |
=== przemeklal_ is now known as przemeklal | ||
teward | cpaelzer: ping | 14:13 |
RoyK | qman__: it's not impossible to bind to a single port when the source port is different | 14:18 |
RoyK | qman__: otherwise, large webservers would be all singlethreaded | 14:20 |
qman__ | binding to a listening port doesn't specify a remote source | 14:20 |
RoyK | qman__: bind() ; fork() | 14:21 |
qman__ | the remote side is * | 14:22 |
qman__ | the scenario you describe is not relevant here | 14:22 |
=== dasm_ is now known as dasm | ||
jamespage | smb: about? I have a bcache related question | 15:36 |
jamespage | when the bcache gc thread runs bcache pauses all frontend IOPS until its completed - but I can't tell whether thats tied to the cache device, pausing IO's to all backing devices it supports, or tied to each backing device | 15:37 |
smb | jamespage, about yes, but I could not answer that question from the top of my head. With stacked devices it could also be the throttling mechanism that somehow tries to limit IO in when the backing device is slow | 15:46 |
ThothCastel | for a home/lab network where linux/mac/win10pro are present on the same network.... to make use of 'LDAP like' authentication, is Ubuntu Server best for that or is it best to run Samba 4 as opposed to enabling the 'ldap server' side of ubuntu? | 17:04 |
ThothCastel | or actually is ubuntu domain controller based on samba 4? | 17:08 |
qman__ | ThothCastel: if you want proper windows support, you need a samba 4 AD DC - I'm not sure what that option does, exactly, there's a few different possibilities | 22:35 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!