Bug 137216
| Summary: | sendmail drops outgoing mail - even if configuration seems correct | ||
|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | carlos puchol <cpg> |
| Component: | sendmail | Assignee: | Thomas Woerner <twoerner> |
| Status: | CLOSED NOTABUG | QA Contact: | |
| Severity: | high | Docs Contact: | |
| Priority: | medium | ||
| Version: | 3 | CC: | cpg |
| Target Milestone: | --- | ||
| Target Release: | --- | ||
| Hardware: | x86_64 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2004-10-28 05:43:36 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
note, to be more precise,
the only thing different than standard config is something that
most people downloading email will have to do and that is:
define(`SMART_HOST',`smtp.comcast.net')
MASQUERADE_DOMAIN(myowndomain.com)dnl
MASQUERADE_AS(myowndomain.com)dnl
hi alexander, Alexander Dalloz wrote: > Am Do, den 28.10.2004 schrieb Carlos Puchol um 1:09: > > Oct 25 14:25:40 rome sendmail[9106]: NOQUEUE: SYSERR(cpg): can not chdir(/var/spool/clientmqueue/): +Permission denied > > > noone running into this? > > No. A test with "date | mail -s test recipient" succeeded without a > problem. > > > -c > > With which MUA do you try to send mail? Seems the mail client running as > user cpg tries to directly feed the submission queue. So it is a task to > configure your client properly. i use mutt. from what you say, i am guessing that the cpg in SYSERR(cpg) means that the id under which it is run is cpg. in that case, the issue is understandable that the process cannot go into the clientmqueue. > Or which permissions does the /usr/sbin/sendmail.sendmail binary have? interesting that you ask that. turns out that it was actually owned by cpg.cpg. hmmm. not sure how that happened, since i typically do all my work under either sudo or su. at any rate, i changed it to smmsp.smmsp and tried again, but got the same issue. furthermore, i got it with the example you mention: [22:12:50](2)rome:~$ date | mail -s test myuser WARNING: RunAsUser for MSP ignored, check group ids (egid=100, want=51) can not chdir(/var/spool/clientmqueue/): Permission denied Program mode requires special privileges, e.g., root or TrustedUser. [22:15:30](2)rome:~$ that clearly indicates that the user id _is_ the issue. [22:23:38](2)rome:~$ id cpg uid=500(cpg) gid=100(users) groups=100(users) [22:23:43](2)rome:~$ id smmsp uid=51(smmsp) gid=51(smmsp) groups=51(smmsp) [22:23:44](2)rome:~$ i removed the rpm, but i took a look at the permissions before doing that: 22:24:38](3)rome:cpg# ll -L /usr/sbin/sendmail* -rwxr-xr-x 1 smmsp smmsp 748296 Sep 1 03:20 /usr/sbin/sendmail -rwsr-xr-x 1 root root 2089729 Sep 13 09:20 /usr/sbin/sendmail.exim -rwxr-xr-x 1 root root 141344 Aug 5 08:03 /usr/sbin/sendmail.postfix -rwxr-xr-x 1 smmsp smmsp 748296 Sep 1 03:20 /usr/sbin/sendmail.sendmail [22:24:42](3)rome:cpg# and the reinstalled. i got this: [22:27:55](3)rome:cpg# ll -L /usr/sbin/sendmail* -rwxr-sr-x 1 root smmsp 748296 Sep 1 03:20 /usr/sbin/sendmail -rwsr-xr-x 1 root root 2089729 Sep 13 09:20 /usr/sbin/sendmail.exim -rwxr-xr-x 1 root root 141344 Aug 5 08:03 /usr/sbin/sendmail.postfix -rwxr-sr-x 1 root smmsp 748296 Sep 1 03:20 /usr/sbin/sendmail.sendmail [22:28:01](3)rome:cpg# which clearly _would_ make things work. if it weren't for the fact that the settings (masquerade and smart host) also reset. then i set them up [22:31:12](3)rome:mail# diff sendmail.mc.rpmsave sendmail.mc 22c22 < define(`SMART_HOST',`smtp.comcast.net') --- > dnl define(`SMART_HOST',`smtp.your.provider') 153c153 < FEATURE(masquerade_envelope)dnl --- > dnl FEATURE(masquerade_envelope)dnl 159,160c159 < MASQUERADE_DOMAIN(mydomain.com)dnl < MASQUERADE_AS(mydomain.com)dnl --- > dnl MASQUERADE_DOMAIN(localhost)dnl [22:31:33](3)rome:mail# cp sendmail.mc.rpmsave sendmail.mc cp: overwrite `sendmail.mc'? y [22:31:37](3)rome:mail# then restarted sendmail a couple of times from the command line and also from the services GUI, in case i could reproduce it. i can't. so -- i am baffled as to how the user/group and the setuid was changed. i am 100% certain i did not change the permissions or user/group by hand. thanks for your help, even though this will remain a mystery. if nothing else, this prompted me to finally switch to postfix, at least for outbound mail. -c |
Description of problem: outgoing sendmail on an fc3t3 stock system with minor changes drops email on the floor (losing data), even as the configuration seems correct. Version-Release number of selected component (if applicable): sendmail-8.13.1-2 How reproducible: 100% on my system Steps to Reproduce: 1. default system 2. /etc/mail/sendmail.mc has - smart host definedm and - masquerade (as my domain) different from base config (but this "should not matter"). 3. send email from mutt Actual results: mo mail is ever sent. Expected results: mail should go out. Additional info: i put one fct3 machine into near "production." i noticed that sendmail dropped outgoing messages on the floor. the error is as follows: Oct 25 14:25:40 rome sendmail[9106]: NOQUEUE: SYSERR(cpg): can not chdir(/var/spool/clientmqueue/): Permission denied now - at this time, i checked and sendmail (the outgoing process) was running as user smmsp. the /var/spool/clientmqueue/ directory had these permissions. drwxrwx--- 2 smmsp smmsp 4096 Oct 26 00:53 /var/spool/clientmqueue/ which should have been fine, according to my research, these permissions should be the correct ones, and the owner/group as well is the correct one (much more so when the process is running with that user id). at any rate, i changed the permissions for that directory to 777 and things started working - mail would go out as expected. i notice that now there is a process that nicely warns that these permissions are (very appropriately!): Oct 25 15:23:12 rome sm-msp-queue[2375]: dangerous permissions=40777 on queue directory /var/spool/clientmqueue/ since there is data loss - and this is the default configuration, this seems like a potential issue (thus i set it as priority high).