[04:20] how to install postfix in ubuntu, I use this guide [04:20] https://help.ubuntu.com/lts/serverguide/postfix.html [04:39] good morning [05:07] cpaelzer: good [05:07] post fix send message and accept, but postfixadmin gets nothing [05:08] maybe need to configure postfix with mysql [05:10] I've tried a few mails to send on gmail and they didn't get to spam. it's nice [05:10] what I need next? [05:11] certificate? [05:11] SASL? [05:11] or ok, better maybe install mysql [05:24] I've got this error: [05:24] neo@v127722:~$ sudo apt-get postfix-mysql [05:24] E: Invalid operation postfix-mysql [05:24] oh [05:24] install [05:26] yep [05:50] what do next? [05:50] I installed postfix with mysql [12:04] who know why dovecot doesn't run? [12:05] I've configured it using this instruction [12:05] https://www.rosehosting.com/blog/set-up-a-mail-server-with-postfixadmin-and-mariadb-on-centos-7/ [12:05] and it stopped to work :( [12:14] I've got error https://paste.ubuntu.com/p/YrkP397tDG/ [16:31] nacc: would you know why I started seeing this in the samba repo just now? https://pastebin.ubuntu.com/p/wpWzSFPMyP/ [16:31] I've done many samba commits/uploads in the past, even the version that is right now in git, but now is the first time this happens [16:32] maybe a new snap? Or it was reimported? [16:38] ahasenack: looking [16:46] ahasenack: dunno, that directory is defintely empty [16:47] ahasenack: did you see if it was added/emptied in the latest upload? [16:47] nacc: I didn't notice any of that [16:48] nacc: and 4.7.6 was uploaded twice [16:48] ah, no, jus tonce [16:48] but the first 4.7.6 upload came via a debian merge [16:48] meaning, the whole git workflow process was used [16:51] ahasenack: i only see one 4.7.6 upload, what do you mean by first? [16:51] ahasenack: i mean i only see 2:4.7.6+dfsg~ubuntu-0ubuntu1 [16:52] and I corrected myself above [16:52] there is only one 4.7.6 upload [16:52] but it should have hit this problem then, or so I thought [16:52] ahasenack: 'first' came after the correction :) [16:53] I am working on the second upload [16:53] ahasenack: that directory didn't exist in 4.7.4, fyi [16:53] ahasenack: link to the merge MP? [16:53] not done, because I hit this problem and I don't know what to do [16:54] source packages are here: https://launchpad.net/~ahasenack/+archive/ubuntu/samba-kerberos-method-1761737 [16:54] a simple patch [16:55] ahasenack: ... the one that resulted in the last upload? [16:55] hm, let's see if I can find it [16:55] ahasenack: you do an MP normally for the new upload, it just won't match the upload tag (so you can upload tag or not, it doesn't matter) [16:56] nacc: this was the MP: https://code.launchpad.net/~ahasenack/ubuntu/+source/samba/+git/samba/+merge/341418 [16:56] oh, interesting [16:56] tar -czf samba_4.7.6+dfsg.orig.tar.gz --exclude="source4/heimdal/lib/wind/rfc*txt" --exclude="source4/ldap_server/devdocs" --exclude="*chm" samba-4.7.6 [16:57] yes, it's excluded because of dfsg [16:57] but it wasn't [16:57] but that didn't exclude the directory itself? [16:57] it's just empty [16:57] yeah, something seems off in your orig [16:58] (i verified the orig by untarring it manually and it definitely has that directory) [16:58] as to why your git commit didn't notice it, i'm not 100% [16:58] possibly an older git clone? [16:58] could be [17:00] I could repackage it one more time, call it 4.7.6+dfsg~ubuntu1 perhaps [17:00] shucks [17:00] ahasenack: no, you don't want to do that, i don't think [17:00] i mean, yes you can, but i don't think you should [17:03] well, it's fine to ignore when committing, we know it was just doc removal that caused it (rfcs essentially) [17:10] When I commission a node in MAAS, It does not detect the storage. How do I trouble shoot this? [17:14] Hey: best to ask in #maas [17:21] nacc: yeah, I don't know what happened there [17:21] andreas@nsnx:~/x$ tar xzf samba-4.7.7.tar.gz [17:21] andreas@nsnx:~/x$ tar -czf samba_4.7.7+dfsg.orig.tar.gz --exclude="source4/heimdal/lib/wind/rfc*txt" --exclude="source4/ldap_server/devdocs" --exclude="*chm" samba-4.7.7 [17:21] andreas@nsnx:~/x$ tar tvzf samba_4.7.7+dfsg.orig.tar.gz |grep devdocs [17:21] andreas@nsnx:~/x$ [17:21] it just works [17:21] maybe something during sponsoring [17:24] (didn't find a 4.7.6 tarball, used 4.7.7 which has the same devdocs directory, for this test) [17:25] nacc: maybe this was used: --exclude="source4/ldap_server/devdocs/*" [17:25] that leaves an empty decdocs directory [17:25] devdocs* [17:26] dpb1: MAAS is dead channel.. I'm sure its not intentional. [17:29] apparently not [17:32] ahasenack: right, i pasted what you said in the MP, which may or may not have been what the sponsor did [17:33] ahasenack: this is why it's usually good (in the future) to put a hash for the tarball in [17:33] cpaelzer: --^ fyi [17:41] ok, I have a patch, it fixes the crash, but I won't get a confirmation from the reporter until tomorrow [17:41] and tomorrow is final freeze [17:41] do I upload or not [17:41] patch is from upstream, oneliner [18:23] I'm so confused. I have a script that makes a backup then encrypts and uploads it to S3. It works fine when I invoke it with "sudo -u backup /bin/bash /usr/local/bin/backup-mysql.sh varible1 varible2" [18:24] But when systemd invokes it uploads a 0 byte file (otherwise backup file on server exists and is not 0 bytes) [18:24] The service file: https://pastebin.com/Rx2rk48g [18:26] I've checked all of the script varibles (dumped to output) and they're exactly the same [18:29] ProCycle: you'll need a shell for the redirection to work [18:29] ProCycle: I'm not sure you can use shell redirect [18:29] you can't [18:29] """Specifically, redirection using "<", "<<", ">", and ">>", pipes using "|", running programs in the background using "&", and other elements of shell syntax are not supported.""" [18:30] from the 'systemd.service' manfile [18:30] ProCycle: based on that, maybe it would be better to edit the backup-mysql.sh script to accept an argument [18:31] * ProCycle looks up shell redirection [18:32] or use "sh -c" when you can't change the backup script [18:32] Oh, uh what do you mean by shell redirection? in my execstart? [18:32] yes [18:33] Oh wait I get what you're seeing, that's a red herring [18:33] is a placeholder, I have an actual space name there [18:34] https://pastebin.com/QCSTieDa [18:34] ProCycle: could you test with sudo -Hu backup ... ? [18:35] ProCycle: if the S3 key is in your home dir, the backup user wouldn't have access to it [18:36] I'm also not sure you test by forcing /bin/bash as the interpreter but that is another delta with how systemd runs the job [18:36] s/sure/sure why/ [18:36] For some reason it won't run otherwise [18:36] ProCycle: is the file executale? [18:36] But I just did sudo -Hu backup and got the same error [18:36] * sdeziel can't type today [18:37] So it must have something to do with the home directory [18:37] Nothing is stored there though [18:37] ProCycle: if you need to specify /bin/bash it could mean 2 things (I think). 1) the shebang is wrong or 2) the file is not executable [18:38] https://github.com/ProCycleDev/debian-ubuntu-mariadb-backup/blob/master/backup-mysql.sh [18:38] checking [18:39] Oh it wasn't set as executable [18:39] ProCycle: I never used s3cmd but I seems possible that it access some files from your home or an env variable. I'd strace it to know [18:45] Hmm it might have something to do with gpg2 [18:45] Since it uses that to encrypt the file [18:47] Shouldn't matter since it's only using a symetrical cipher but then again gpg2 is very inistent you use the agent even when not using keys [18:48] It breaks all the same if I run the s3cmd manually with sudo -hu backup [18:49] Problem is the backup user is a system user and has no shell login. Probably need to create a normal user instead [18:49] Cannot redirect from http to https, what can be the reason https://pastebin.mozilla.org/9083351 ? [18:52] ProCycle: I use "openssl enc -aes128" for such needs but I think it's not considered very secure [18:53] Sircle, Try doing this instead https://pastebin.com/45ZjSzM3 [18:54] ProCycle, I should do it in :80 vhost and :443 vhost as well? [18:56] In my config I'm not using them, but I'd assume you'd do a vhost for your domain instead of the port [18:56] can you paste your config [18:57] The rewrite rule just redirects them if they're not using HTTPS, otherwise nothing happens and things proceed normally [18:57] That's literally my entire config (in an .htaccess file) [18:57] I mostly use nginx nowadays [18:57] is it inside a vhost of a site or its an open config [18:58] But that snippet is from apache's help site [18:59] I'd make a single vhost for your domain like normal and use that snippet to redirect non HTTPS connections [18:59] can you paste? [19:01] Do you only have one site on this apache install? [19:01] Or are there multiple domains? [19:02] https://httpd.apache.org/docs/2.4/mod/core.html#virtualhost [19:06] So something like https://pastebin.com/wJ15shct [19:06] ProCycle, I have many [19:06] and each is in its vhost [19:06] Except correct for your site setup (I'm having a hard time making sense of your config file) [19:06] Oh so you'd do [19:07] Not sure if that applies to www too (probably not) [19:07] Sorry can't be more help, rather rusty on apache [19:08] I was missing RewriteEngine on [19:40] powersj: ping [19:41] nacc: hey [19:41] powersj: got a quick pytest question for you if have a sec [19:42] nacc: sure [19:42] powersj: could you join the standup HO? I think i can explain it fastest there [19:42] omw [19:43] powersj: thanks [20:12] sdeziel, Yep it was a problem with the user, couldn't write to it's own home directory. I ended up making a new user instead and now it works. Thanks for your insight! [20:50] ProCycle: great [22:50] ProCycle, I have made it simpler. Can you see the error. How can I know the reason behind it? https://pastebin.mozilla.org/9083373 [22:53] I'm sorry that's out of my scope of knowledge about apache [23:10] Sircle, either the proxy portion isnt set up correctly and not working, or the user or path doesnt exist. [23:10] or permissions on the path is wrong [23:11] could be many things [23:13] how can I back trace? [23:14] maybe try tcpdump or tshark and see what is being senton the wire? [23:25] Sircle: first of all, just try to nc -vv to the backend hostname and port, to ensure the tcp port is actually open / listening for cxonnections from this host, [23:29] tomreyn, nc -x localhost:50002 -vv? [23:29] better use the kryptowhatever.com address here too [23:29] k [23:30] whats the actual command? [23:30] Sircle: what sarnold says, and i don'T see port 50002 listend in what you posted so far. the error message is about port 80 [23:31] *lsited [23:31] *listed [23:31] tomreyn: the 50002 is from line 9 https://pastebin.mozilla.org/9083373 [23:31] nmap localhost says 50002/tcp open iiimsf [23:32] sorry i'm blind. but why does line 23 refer to port 80? [23:32] we did not ask about localhost [23:32] I assumed that the was the apache server doing the proxying [23:32] I also assumed that localhost == knockcryptoapi.com [23:33] kryptowhatever.com should not resolve to an ip address which reverse resolves to localhost. [23:34] right [23:34] nor knockcryptoapi [23:34] which is why I suggested using the knockcryptoapi.com dns name in the nc tests [23:34] to make sure it resolves to something that the host itself can reach [23:35] and i agree with this approach [23:35] sarnold, yes. ignore localhost [23:35] so why its not proxying well? [23:38] we don't know and cannot speculate since you seem to be withholding relevant information. We suggest you try to have the system which reports "error reading status line from remote server knockcryptoapi.com:50002" to run "nc -vv knockcryptoapi.com:50002" to ensure that tcp conmmunication between it and its backend works properly. [23:39] typo. this should haver been: "nc -vv knockcryptoapi.com 50002" [23:41] tomreyn, https://pastebin.mozilla.org/9083377 [23:42] oh ok [23:42] no colon :) [23:42] onnection to knockcryptoapi.com 50002 port [tcp/*] succeeded! [23:44] so tcp connectivity works generally. then you'll want to investigate why "AH01102: error reading status line from remote server knockcryptoapi.com:50002" and "AH00898: Error reading from remote server returned by /cryptoarchid/webapi/v1/person/test " were triggered. [23:46] experiment with curl, tcpdump, firefox / chromium network console, burp proxy... whatever you prefer. [23:48] http://knockcryptoapi.com:50002/ is just running fine [23:48] its not proxing [23:49] Status Code: 502 Proxy Error [23:49] then bust out tshark or tcpdump and watch the traffic between the proxy and the :50002 program and try to spot the error? [23:49] Remote Address: 107.161.18.128:80 [23:49] Referrer Policy: no-referrer-when-downgrade [23:49] then do whatever is needed to make it proxy. [23:51] I have made it simpler. Can you see the error. How can I know the reason behind it? https://pastebin.mozilla.org/9083373 [23:52] does the program on :50002 have logs? [23:52] yes [23:57] ok, will do rnd and let you know