Bug #2788
closedtopology alignment script fails shortly before uploading class averages
0%
Description
Reported by Tom Houweling
appion svn r18192 topologyAlignment script fail
This has occurred with two separate data sets with two separate experienced users.
Command:
topologyAlignment.py --projectid=86 --rundir=/k2data/appion/taylor/14may08a/align/topol4 --description="test" --runname=topol4 --stack=1 --lowpass=10 --highpass=2000 --num-part=340 --start=5 --end=19 --bin=2 --mask=70 --iter=5 --itermul=10 --learn=0.01 --ilearn=0.0005 --age=25 --mramethod=imagic --msamethod=can --nproc=8 --msaproc=8 --commit --expid=2403 --jobtype=partalign
Error log leading up to buffer overflow:
calculating class average : 0total number of node images = 19
outputing average images to : classes_avg.hed
total number of avg images = 19
executing IMAGIC command: /opt/qb3/imagic-110817e/stand/arithm.e
!!! WARNING: removing stack: classes.hed
!!! WARNING: removing stack: classes.img
executing IMAGIC command: /opt/qb3/imagic-110817e/stand/pretreat.e
!!! WARNING: removing stack: classes_mask.hed
!!! WARNING: removing stack: classes_mask.img
finished alignment in 9.03 sec
EMAN: proc2d classes_mask_norm.hed /k2data/appion/taylor/14may08a/align/topol4/classes05.hed
!!! WARNING: removing stack: classes_mask_norm.hed
!!! WARNING: removing stack: classes_mask_norm.img
... Alignment time: 2 min 24 sec
... Sorting final class averages
!!! WARNING: replacing stack file '/k2data/appion/taylor/14may08a/align/topol4/classes05.hed' with 'sortedcls.hed'
!!! WARNING: replacing stack file '/k2data/appion/taylor/14may08a/align/topol4/classes05.img' with 'sortedcls.img'
executing IMAGIC command: /opt/qb3/imagic-110817e/stand/testim.e
... replacing 'mrastack.hed' with 'tempimg.hed'
- buffer overflow detected ***: python terminated
======= Backtrace: =========
/lib/libc.so.6(_fortify_fail+0x37)[0x7fe783be9cd7]
/lib/libc.so.6(+0x101b90)[0x7fe783be8b90]
/lib/libc.so.6(+0x100ff9)[0x7fe783be7ff9]
/lib/libc.so.6(_IO_default_xsputn+0xcc)[0x7fe783b5e35c]
/lib/libc.so.6(_IO_vfprintf+0x3b96)[0x7fe783b31286]
/lib/libc.so.6(_vsprintf_chk+0x99)[0x7fe783be8099]
/lib/libc.so.6(_sprintf_chk+0x7f)[0x7fe783be7fdf]
/opt/qb3/eman-2.05/lib/libEM2.so(_ZN4EMAN9ImagicIO212write_headerERKNS_4DictEiPKNS_6RegionENS_6EMUtil10EMDataTypeEb+0x84f)[0x7fe7831d8fef]
/opt/qb3/eman-2.05/lib/libEM2.so(_ZN4EMAN6EMData11write_imageERKSsiNS_6EMUtil9ImageTypeEbPKNS_6RegionENS3_10EMDataTypeEb+0x6d5)[0x7fe78313b555]
/opt/qb3/eman-2.05/lib/libpyEMData2.so(+0xb4360)[0x7fe77fc6e360]
/opt/qb3/eman-2.05/lib/libpyEMData2.so(_ZN5boost6python7objects23caller_py_function_implINS0_6detail6callerIPFvRN4EMAN6EMDataERKSsiENS0_21default_call_policiesENS_3mpl7vector4IvS7_S9_iEEEEEclEP7_objectSJ+0x128)[0x7fe77ff48da8]
Updated by Anchi Cheng over 10 years ago
- Status changed from New to Assigned
- Assignee set to Gabriel Lander
- Priority changed from Normal to High
Tom,
Please update this issue with how many items were in the stack mrastack that caused the error. If it does not happen all the time, the size might matter.
Gabe,
I think this is related to your code. Please take a look. Thanks.
Updated by Gabriel Lander over 10 years ago
- Status changed from Assigned to New
- Assignee deleted (
Gabriel Lander) - Priority changed from High to Normal
This may be occurring during the "sortClassAverages" function in topologyAlignment.py, which uses EMAN2 to sort the class averages by similarity. I've never seen this error though - maybe an incompatibility between IMAGIC and EMAN2?
At our end we use imagic/100312 & EMAN2.04, and we don't have problems. I'm pretty sure we've tried EMAN2.06 and this works as well.
Can you try out different versions of imagic & EMAN2 to see if you always get the same error?
Updated by Gabriel Lander over 10 years ago
- Assignee set to Tom Houweling
Fixed the incompatibility with the "classes_avg" imagic-formatted output experienced at Berkeley by using EMAN1 to convert the stack with the command:
"proc2d fname fname inplace"
This should be fixed in svn r18289
Tom could you update topology alignment, test and close the issue if it's working?
Updated by Anchi Cheng over 10 years ago
Hi, Tom,
This is to reply your e-mail so that you can close this issue.
With Redmine email sent through wwwami, you should find an url link to the issue in our reamine webpage. Go to that url shows you the issue with the progress so far. If you log in as youself, you can then update the issue with a change of the status or add your comments. When you submit the update, all watchers (Amber,Gabe, you and me) will then get the update text you submit and the status change like this update I am writing. By default the sender does not get the e-mail.
Since Gabe has already updated the code in r18289, if your change is the same, you can just close the issue. Otherwise, test his change first.
Updated by Tom Houweling over 10 years ago
- Status changed from New to Closed
- Affected Version changed from Appion/Leginon 2.1.0 to Appion/Leginon 3.0.0
Tested the fix with rev 18289. Problem is fixed.