When I try to update QuasiLocalMeasures from LSUThorns, I receive the following error message:
Updating module: LSUThorns/QuasiLocalMeasures from repository: :pserver:cvs.cct.lsu.edu/numrel located in: Cactus/arrangements cvs [update aborted]: received broken pipe signal
ERROR: Could not update module LSUThorns/QuasiLocalMeasures
Died at Utilities/Scripts/GetComponents line 1402.
There seem to be two problems: - CVS at the CCT is broken - GetComponents should not abort if it encounters an errors; instead, it should skip the module (or the repository?) and report the error message later
-erik
Hello Erik,
When I try to update QuasiLocalMeasures from LSUThorns, I receive the following error message:
Updating module: LSUThorns/QuasiLocalMeasures from repository: :pserver:cvs.cct.lsu.edu/numrel located in: Cactus/arrangements cvs [update aborted]: received broken pipe signal
ERROR: Could not update module LSUThorns/QuasiLocalMeasures
Died at Utilities/Scripts/GetComponents line 1402.
There seem to be two problems:
- CVS at the CCT is broken
- GetComponents should not abort if it encounters an errors; instead,
it should skip the module (or the repository?) and report the error message later
Does this happen on a machine within LSU or outside of LSU? Neither Bruno nor I could ever *update* QLM from outside of LSU. I have a ticket open with computer support and could narrow it down somewhat but have not heard from support in a while.
Yours, Roland
I have had similar issues with the LSUThorns on teragrid machines, although it worked fine on my laptop in NYC. I think the die on error part of GetComponents was put in for testing purposes; I can remove it when I get back to my computer.
Eric
Von meinem iPhone gesendet
Am May 30, 2010 um 11:58 schrieb Roland Haas <roland.haas@physics.gatech.edu
:
Hello Erik,
When I try to update QuasiLocalMeasures from LSUThorns, I receive the following error message:
Updating module: LSUThorns/QuasiLocalMeasures from repository: :pserver:cvs.cct.lsu.edu/numrel located in: Cactus/arrangements cvs [update aborted]: received broken pipe signal
ERROR: Could not update module LSUThorns/QuasiLocalMeasures
Died at Utilities/Scripts/GetComponents line 1402.
There seem to be two problems:
- CVS at the CCT is broken
- GetComponents should not abort if it encounters an errors; instead,
it should skip the module (or the repository?) and report the error message later
Does this happen on a machine within LSU or outside of LSU? Neither Bruno nor I could ever *update* QLM from outside of LSU. I have a ticket open with computer support and could narrow it down somewhat but have not heard from support in a while.
Yours, Roland
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
On May 30, 2010, at 11:58 , Roland Haas wrote:
Hello Erik,
When I try to update QuasiLocalMeasures from LSUThorns, I receive the following error message:
Updating module: LSUThorns/QuasiLocalMeasures from repository: :pserver:cvs.cct.lsu.edu/numrel located in: Cactus/arrangements cvs [update aborted]: received broken pipe signal
ERROR: Could not update module LSUThorns/QuasiLocalMeasures
Died at Utilities/Scripts/GetComponents line 1402.
There seem to be two problems:
- CVS at the CCT is broken
- GetComponents should not abort if it encounters an errors; instead,
it should skip the module (or the repository?) and report the error message later
Does this happen on a machine within LSU or outside of LSU? Neither Bruno nor I could ever *update* QLM from outside of LSU. I have a ticket open with computer support and could narrow it down somewhat but have not heard from support in a while.
This is from home. I think it works fine from LSU. Could IT support reproduce the problem? If not, they probably won't be able to solve it...
-erik
On 31/05/2010 04:48, Erik Schnetter wrote:
This is from home. I think it works fine from LSU. Could IT support reproduce the problem? If not, they probably won't be able to solve it...
I am also experiencing this problem from the AEI. I am able to checkout the repository but can't do a cvs update, cvs diff or cvs status. Attached is a dump of the network packets when I tried to run cvs status. This is in the standard pcap format. You can see that the cvs client is sending the request:
BEGIN AUTH REQUEST./numrel.cvs_anon.Ay=0=.END AUTH REQUEST.
and is not getting a response from the server. So, either the server is not responding to this request or the LSU firewall is blocking the packets for computers outside the LSU network. Is this any help in narrowing down the problem?
Regards, Barry
Hello Erik, Barry,
This is from home. I think it works fine from LSU.
It worked for me from within LSU, yes (on is.cct.lsu.edu). Just cvs update will do the trick, GetComponents is not required.
Could IT support reproduce the problem? If not, they probably won't be able to solve it...
They never seem to have tried. maybe they do not have access to an outside machine?
I me sending emails back and forth with them I could narrow the problem down to find one file that gives trouble.
/LSUThorns/QuasiLocalMeasures/test/qlm-minkowski/quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc
Which I cannot update (but can check out).
I have enclosed a transcript of the latest email. I have not heard from support since then (neither that they have given up nor that the problem has been solved).
Yours, Roland
Hello Dilip,
Please try updating your local copy. If it doesn't work and still it gives you an error, try deleting the local copy of the file and try cvs update. If it fails let me know.
Ok I do: [rhaas3@phys44230 qlm-minkowski]$ cvs update quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc ..stalls.. cvs [update aborted]: received interrupt signal [rhaas3@phys44230 qlm-minkowski]$ rm quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc [rhaas3@phys44230 qlm-minkowski]$ cvs update quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc ..stalls.. cvs [update aborted]: received interrupt signal
it still stalls both before and after I remove the local copy (I press CTRL-C in both cases eventually). Just to make sure I was not to hasty aborting I had another update run since the last email I send to you, it eventually timed out with:
[rhaas3@phys44230 qlm-minkowski]$ cvs update quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc cvs [update aborted]: reading from server: Connection timed out
Other files seem to be fine: [rhaas3@phys44230 qlm-minkowski]$ rm carpetreduce::weight.z.asc [rhaas3@phys44230 qlm-minkowski]$ cvs update carpetreduce::weight.z.asc cvs update: warning: carpetreduce::weight.z.asc was lost U carpetreduce::weight.z.asc
What also worked was:
[rhaas3@phys44230 qlm-minkowski]$ cd .. [rhaas3@phys44230 test]$ rm qlm-minkowski -rf [rhaas3@phys44230 test]$ cvs update -CPd qlm-minkowski
doing another [rhaas3@phys44230 test]$ cvs update -CPd qlm-minkowski
right afterwards hangs again.
Yours, Roland
Roland Haas ha scritto:
I me sending emails back and forth with them I could narrow the problem down to find one file that gives trouble.
/LSUThorns/QuasiLocalMeasures/test/qlm-minkowski/quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc
Which I cannot update (but can check out).
I have enclosed a transcript of the latest email. I have not heard from support since then (neither that they have given up nor that the problem has been solved).
Hi all,
has this issue been resolved in the end? It appears that I'm having the same problem from damiana, but in an intermittent and unreproducible way.
Thanks, Eloisa
On 14 Jun 2010, at 14:36, Eloisa Bentivegna wrote:
Roland Haas ha scritto:
I me sending emails back and forth with them I could narrow the problem down to find one file that gives trouble.
/LSUThorns/QuasiLocalMeasures/test/qlm-minkowski/ quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc
Which I cannot update (but can check out).
I have enclosed a transcript of the latest email. I have not heard from support since then (neither that they have given up nor that the problem has been solved).
Hi all,
has this issue been resolved in the end? It appears that I'm having the same problem from damiana, but in an intermittent and unreproducible way.
I am also having the problem - I attempted to do a checkout using GetComponents including the thorn LSUThorns/SummationByParts, and it hung on the file stencil.h (the last file, though I don't know if this is important). When I went into the directory and did a cvs update, it fetched the file and all was well. This was from within the AEI network.
Hello Eloisa,
has this issue been resolved in the end? It appears that I'm having the same problem from damiana, but in an intermittent and unreproducible way.
I can now to a cvs update without a hang. However the file that was giving me trouble
/LSUThorns/QuasiLocalMeasures/test/qlm-minkowski/quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc
is also no longer present. I have still not heard from cct support, so have no idea whether they were involved or not.
Yours, Roland
My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from http://keys.gnupg.net.
On Jun 14, 2010, at 8:00 , Roland Haas wrote:
Hello Eloisa,
has this issue been resolved in the end? It appears that I'm having the same problem from damiana, but in an intermittent and unreproducible way.
I can now to a cvs update without a hang. However the file that was giving me trouble
/LSUThorns/QuasiLocalMeasures/test/qlm-minkowski/ quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc
is also no longer present. I have still not heard from cct support, so have no idea whether they were involved or not.
I found that I could solve the problem for myself (for using my notebook from home) by adding the "-z9" option to cvs. I added this to GetComponents.
Does the problem persist for you even with this change?
-erik
Hello Erik,
I found that I could solve the problem for myself (for using my notebook from home) by adding the "-z9" option to cvs. I added this to GetComponents.
I noticed that commit (and the commit message). I am not sure if this really solves the problem or just makes it less likely to appear because the amount of data transferred is smalller.
Does the problem persist for you even with this change?
Anyway I did a
cvs -d:pserver:cvs_anon:anon@cvs.cct.lsu.edu/numrel login cvs -d:pserver:cvs_anon:anon@cvs.cct.lsu.edu/numrel co LSUThorns/QuasiLocalMeasures cd LSUThorns/ cvs update
which worked fine. However as said, the file that was causing the hang
/LSUThorns/QuasiLocalMeasures/test/qlm-minkowski/quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc
is now gone from the repository. In fact cvs claims that the file never existed in the first place (tested from a checkout within LSU, see below). Did you ask the cvs admins to permanently delete some files from the repository?
Also if I do a 'cvs log' in QuasiLocalMeasures then cvs still seems to hang. cd QuasiLocalMeasures cvs log & ..wait..several..minutes.. netstat --tcp | grep 'lsu|^Prot
Proto Recv-Q Send-Q Local Address Foreign Address State tcp 0 43776 phys44230.physics.gat:42014 asylum.cct.lsu.e:cvspserver ESTABLISHED
So the error seems to still persist. cvs log from is.cct.lsu.edu works fine. Since svn seems to be fine, would maybe moving LSUThorns to svn be a workaround?
Yours, Roland
On Jun 14, 2010, at 21:23 , Roland Haas wrote:
Hello Erik,
I found that I could solve the problem for myself (for using my notebook from home) by adding the "-z9" option to cvs. I added this to GetComponents.
I noticed that commit (and the commit message). I am not sure if this really solves the problem or just makes it less likely to appear because the amount of data transferred is smalller.
Does the problem persist for you even with this change?
Anyway I did a
cvs -d:pserver:cvs_anon:anon@cvs.cct.lsu.edu/numrel login cvs -d:pserver:cvs_anon:anon@cvs.cct.lsu.edu/numrel co LSUThorns/ QuasiLocalMeasures cd LSUThorns/ cvs update
which worked fine. However as said, the file that was causing the hang
/LSUThorns/QuasiLocalMeasures/test/qlm-minkowski/ quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc
is now gone from the repository. In fact cvs claims that the file never existed in the first place (tested from a checkout within LSU, see below). Did you ask the cvs admins to permanently delete some files from the repository?
No, I did not.
Also if I do a 'cvs log' in QuasiLocalMeasures then cvs still seems to hang. cd QuasiLocalMeasures cvs log & ..wait..several..minutes.. netstat --tcp | grep 'lsu|^Prot
Proto Recv-Q Send-Q Local Address Foreign Address State tcp 0 43776 phys44230.physics.gat:42014 asylum.cct.lsu.e:cvspserver ESTABLISHED
So the error seems to still persist. cvs log from is.cct.lsu.edu works fine. Since svn seems to be fine, would maybe moving LSUThorns to svn be a workaround?
This would probably not only work around, but also solve a few other problems.
-erik
Erik Schnetter ha scritto:
On Jun 14, 2010, at 21:23 , Roland Haas wrote:
However as said, the file that was causing the hang
/LSUThorns/QuasiLocalMeasures/test/qlm-minkowski/ quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc
is now gone from the repository. In fact cvs claims that the file never existed in the first place (tested from a checkout within LSU, see below). Did you ask the cvs admins to permanently delete some files from the repository?
No, I did not.
Roland,
can you be referring to the file quasilocalmeasures::qlm_invariant_coordinates.norm_inf2.asc instead? This *is* in the repository, although I've just replace it with quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc in the last output-file commit since the 2 is obviously a typo (I'm not aware of such a reduction handle, but most importantly that is not what was requested in the parameter file). However, I doubt this problem has anything to do with the hang-ups, since the repository can be accessed fine from LSU machines.
Eloisa
Hello Eloisa,
can you be referring to the file quasilocalmeasures::qlm_invariant_coordinates.norm_inf2.asc instead?
I checked my initial email to ccthelpdesk (Dilip Somavarapu) where I had copied and pasted the cvs output and there apparently used to be a file quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc (this was on May 18th). Also if I run the test then it creates such a file (for the norm_inf reduction), so I suspect that maybe one of the admins renamed the quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc,v file for testing purposes?
This *is* in the repository, although I've just replace it with quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc in the last output-file commit since the 2 is obviously a typo (I'm not aware of such a reduction handle, but most importantly that is not what was requested in the parameter file). However, I doubt this problem has anything to do with the hang-ups, since the repository can be accessed fine from LSU machines.
Well quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc was the file where cvs hung, so it seems somehow related even if I agree that is likely not the actual cause. I had originally done
[rhaas3@phys44230 qlm-minkowski]$ for i in * ; do echo $i ; cvs update $i ; done BH_diagnostics.ah1.gp CVS cvs update: warning: directory CVS specified in argument cvs update: but CVS uses CVS for its own purposes; skipping CVS directory cvs update: warning: directory CVS specified in argument cvs update: but CVS uses CVS for its own purposes; skipping CVS directory admbase::curv.average.asc admbase::curv.count.asc admbase::curv.maximum.asc admbase::curv.minimum.asc admbase::curv.norm1.asc admbase::curv.norm2.asc admbase::curv.norm_inf.asc admbase::lapse.average.asc admbase::lapse.count.asc admbase::lapse.maximum.asc admbase::lapse.minimum.asc admbase::lapse.norm1.asc admbase::lapse.norm2.asc admbase::lapse.norm_inf.asc admbase::metric.average.asc admbase::metric.count.asc admbase::metric.maximum.asc admbase::metric.minimum.asc admbase::metric.norm1.asc admbase::metric.norm2.asc admbase::metric.norm_inf.asc admbase::shift.average.asc admbase::shift.count.asc admbase::shift.maximum.asc admbase::shift.minimum.asc admbase::shift.norm1.asc admbase::shift.norm2.asc admbase::shift.norm_inf.asc carpetreduce::weight..asc carpetreduce::weight.average.asc carpetreduce::weight.count.asc carpetreduce::weight.d.asc carpetreduce::weight.maximum.asc carpetreduce::weight.minimum.asc carpetreduce::weight.norm1.asc carpetreduce::weight.norm2.asc carpetreduce::weight.norm_inf.asc carpetreduce::weight.x.asc carpetreduce::weight.y.asc carpetreduce::weight.z.asc quasilocalmeasures::qlm_3determinant.average.asc quasilocalmeasures::qlm_3determinant.count.asc quasilocalmeasures::qlm_3determinant.maximum.asc quasilocalmeasures::qlm_3determinant.minimum.asc quasilocalmeasures::qlm_3determinant.norm1.asc quasilocalmeasures::qlm_3determinant.norm2.asc quasilocalmeasures::qlm_3determinant.norm_inf.asc quasilocalmeasures::qlm_3determinant.x.asc quasilocalmeasures::qlm_3determinant.y.asc quasilocalmeasures::qlm_3determinant.z.asc quasilocalmeasures::qlm_coordinates.average.asc quasilocalmeasures::qlm_coordinates.count.asc quasilocalmeasures::qlm_coordinates.maximum.asc quasilocalmeasures::qlm_coordinates.minimum.asc quasilocalmeasures::qlm_coordinates.norm1.asc quasilocalmeasures::qlm_coordinates.norm2.asc quasilocalmeasures::qlm_coordinates.norm_inf.asc quasilocalmeasures::qlm_coordinates.x.asc quasilocalmeasures::qlm_coordinates.y.asc quasilocalmeasures::qlm_coordinates.z.asc quasilocalmeasures::qlm_grid_int..asc quasilocalmeasures::qlm_grid_real..asc quasilocalmeasures::qlm_invariant_coordinates.average.asc quasilocalmeasures::qlm_invariant_coordinates.count.asc quasilocalmeasures::qlm_invariant_coordinates.maximum.asc quasilocalmeasures::qlm_invariant_coordinates.minimum.asc quasilocalmeasures::qlm_invariant_coordinates.norm1.asc quasilocalmeasures::qlm_invariant_coordinates.norm2.asc quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc ..stalls.. cvs [update aborted]: received interrupt signal
Maybe some timeout for the server is too small and this only shows up from outside?
By now I have more or less given up on this CVS repository since we do not use the thorn here even though it seems useful :-(
Yours, Roland
On Jun 15, 2010, at 7:28 , Roland Haas wrote:
Hello Eloisa,
can you be referring to the file quasilocalmeasures::qlm_invariant_coordinates.norm_inf2.asc instead?
I checked my initial email to ccthelpdesk (Dilip Somavarapu) where I had copied and pasted the cvs output and there apparently used to be a file quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc (this was on May 18th). Also if I run the test then it creates such a file (for the norm_inf reduction), so I suspect that maybe one of the admins renamed the quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc,v file for testing purposes?
This is getting strange.
Roland, did you say it was your ticket with CCT support? Do you want to call them and see what is happening? It is not good if files get renamed -- either the disk is having problems, or they forgot to rename it back, which could lead to all sorts of trouble for us (imagine this had been a source file -- what happens to its history?)
-erik
Roland Haas ha scritto:
Hello Eloisa,
can you be referring to the file quasilocalmeasures::qlm_invariant_coordinates.norm_inf2.asc instead?
I checked my initial email to ccthelpdesk (Dilip Somavarapu) where I had copied and pasted the cvs output and there apparently used to be a file quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc (this was on May 18th). Also if I run the test then it creates such a file (for the norm_inf reduction), so I suspect that maybe one of the admins renamed the quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc,v file for testing purposes?
Yes, that sounds very plausible.
[rhaas3@phys44230 qlm-minkowski]$ for i in * ; do echo $i ; cvs update $i ; done BH_diagnostics.ah1.gp ... quasilocalmeasures::qlm_invariant_coordinates.norm_inf.asc ..stalls.. cvs [update aborted]: received interrupt signal
Maybe some timeout for the server is too small and this only shows up from outside?
That should be quite easy to test from the admins' end. It is also consistent with Erik's solution of changing the compression level, and with the reported hang-ups while checking out other LSUThorns parts, like SummationsByParts.
By now I have more or less given up on this CVS repository since we do not use the thorn here even though it seems useful :-(
Can I still ask you to add one last message to the open ticket, with (i) a question about the origin of quasilocalmeasures::qlm_invariant_coordinates.norm_inf2.asc and (ii) a suggestion that the server may be timing out? I suspect we'll wind up moving the thorn to SVN very soon anyway - but it would be good to understand what's going on.
Thanks! Eloisa
users@lists.einsteintoolkit.org
-
Barry Wardell -
Eloisa Bentivegna -
Eric Seidel -
Erik Schnetter -
Ian Hinder -
Roland Haas