getent: handle errors

[faxm0dem]

I'm stumbling on the following error message for large groups:

syslog-ng[9522]: $(getent group) failed; key='124', errno='Numerical result out of range (34)'

Found some references:

• http://en.it-usenet.org/thread/18727/14315/ (BROKEN LINK)
• http://www.spinics.net/lists/hotplug/msg02552.html
• https://fedorahosted.org/sssd/ticket/153

[faxm0dem]

Trying to solve this with my limited Cskills.
I'm seeing this error for groups with the following sizes (according to getent group <gid> | wc -c): 875, 726, 1004, ... . Looking at the code, I see bufsize which on my system seems to be 1024: shouldn't that be enough for these groups?

[faxm0dem]

I guess the solution would be to have a loop reallocating the buffer in case if (errno == ERANGE)

[algernon]

As a quick workaround, I'd try upping the buffer size to something like 4096. That's easier and less costy than reallocation would be.

[faxm0dem]

you mean by ignoring _SC_GETPW_R_SIZE_MAX completely?

[algernon]

Yes.

[mochrul]

getent module removed from the incubator (and merged into syslog-ng upstream).