Hello,
we've installed Zimbra 8.0 NE on Ubuntu 12.04 LTS 64bit. Now I'v added some useraccounts and imported some mails (via imapsync from our old Scalix server). Everything seems to rune fine, but I get a lot of errors about an undefined symbol in pdfsr.so in /opt/zimbra/log/convertd.log....:
...
/opt/zimbra/httpd-2.4.2/bin/httpd: symbol lookup error: /opt/zimbra/keyview/FilterSDK/bin/pdfsr.so: undefined symbol: atan
[Tue Oct 23 17:22:59.450987 2012] [:info] [pid 954] mod_convert: clean dir=/opt/zimbra/convertd/convert/convert
[Tue Oct 23 17:22:59.451355 2012] [:info] [pid 954] mod_convert: clean dir=/opt/zimbra/data/tmp/convert
/opt/zimbra/httpd-2.4.2/bin/httpd: symbol lookup error: /opt/zimbra/keyview/FilterSDK/bin/pdfsr.so: undefined symbol: atan
[Tue Oct 23 17:22:59.743815 2012] [:info] [pid 987] mod_convert: clean dir=/opt/zimbra/convertd/convert/convert
[Tue Oct 23 17:22:59.744144 2012] [:info] [pid 987] mod_convert: clean dir=/opt/zimbra/data/tmp/convert
/opt/zimbra/httpd-2.4.2/bin/httpd: symbol lookup error: /opt/zimbra/keyview/FilterSDK/bin/pdfsr.so: undefined symbol: atan
[Tue Oct 23 17:23:01.853784 2012] [:info] [pid 1025] mod_convert: clean dir=/opt/zimbra/convertd/convert/convert
[Tue Oct 23 17:23:01.854153 2012] [:info] [pid 1025] mod_convert: clean dir=/opt/zimbra/data/tmp/convert
...
should I worry about this? Any known fix?
Walter
symbol lookup errors in convertd.log
Re: symbol lookup errors in convertd.log
8 Year bump. I've just noticed this on an 8.8.15_p9 and an 9.0.0 install.
zimbra@zimbra:~/log$ zmcontrol -v
Release 9.0.0.GA.3924.UBUNTU18.64 UBUNTU18_64 NETWORK edition, Patch 9.0.0_P2.
From what I can see, pdfsr.so gets atan from libc6 and there isn't an issue finding it. Given the keyview binaries were compiled in 2011, I wonder if there's a subtle change in the atan calling format.
A few bugs about it in the bugzilla, and all on Ubuntu, but no resolutions.
zimbra@zimbra:~/log$ zmcontrol -v
Release 9.0.0.GA.3924.UBUNTU18.64 UBUNTU18_64 NETWORK edition, Patch 9.0.0_P2.
Code: Select all
zimbra@zimbra:~/log$ cat convertd.log.2020-05-15
[Fri May 15 09:35:40.926152 2020] [:info] [pid 28374] mod_convert: clean dir=/opt/zimbra/convertd/convert/convert
[Fri May 15 09:35:40.926288 2020] [:info] [pid 28374] mod_convert: clean dir=/opt/zimbra/data/tmp/convert
/opt/zimbra/common/bin/httpd: symbol lookup error: /opt/zimbra/keyview/FilterSDK/bin/pdfsr.so: undefined symbol: atan
[Fri May 15 09:35:57.072945 2020] [:info] [pid 28375] mod_convert: clean dir=/opt/zimbra/convertd/convert/convert
[Fri May 15 09:35:57.073127 2020] [:info] [pid 28375] mod_convert: clean dir=/opt/zimbra/data/tmp/convert
A few bugs about it in the bugzilla, and all on Ubuntu, but no resolutions.
Code: Select all
zimbra@zimbra:~/log$ ldd /opt/zimbra/keyview/FilterSDK/bin/pdfsr.so
linux-vdso.so.1 (0x00007ffc2d3fc000)
libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007f3190ce3000)
libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f31908f2000)
libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007f31906d3000)
/lib64/ld-linux-x86-64.so.2 (0x00007f3191100000)
Re: symbol lookup errors in convertd.log
I'm pleased you got it working, although I can't figure out either what you got working or how it's relevant to the problem in the thread.TerranceM wrote:Finally I installed blasr 5.3.1. For the new error: "blasr: error while loading shared libraries: libpbbam.so.0.18.0: cannot open shared object file: No such file or directory", I just installed that specific version of pbbam. Some additional changes were automatically introduced during installation, but finally it worked. So far so good.
Re: symbol lookup errors in convertd.log
G'day,
This has been an ongoing issue but I just stumbled across this Wiki post from 2021 with the solution : https://wiki.zimbra.com/wiki/Fix_Search ... ttachments
Quote : "Search and indexing with the PDF files are broken because Convertd is unable to parse and index PDF files on Ubuntu OS."
Am I the only one who thinks it's odd that Zimbra has a bug in the Ubuntu version, and obviously knows about the bug. Rather than actually fix the bug, just post a wiki post saying "here's how to fix the the bug" and do it in a manner where you'll have to re-apply the fix any time the zimbra-convertd packaged is updated.
Anyway, at least I have a workaround.
This has been an ongoing issue but I just stumbled across this Wiki post from 2021 with the solution : https://wiki.zimbra.com/wiki/Fix_Search ... ttachments
Quote : "Search and indexing with the PDF files are broken because Convertd is unable to parse and index PDF files on Ubuntu OS."
Am I the only one who thinks it's odd that Zimbra has a bug in the Ubuntu version, and obviously knows about the bug. Rather than actually fix the bug, just post a wiki post saying "here's how to fix the the bug" and do it in a manner where you'll have to re-apply the fix any time the zimbra-convertd packaged is updated.
Anyway, at least I have a workaround.