Re: [vchkpw] vdeluser isCatchall mysql segfault (was Re: [vchkpw] Decided to make the plunge today on late version of vpopmail)
Marcello Lupo wrote: Hi, i agree with your method to bybass the problem but in this way you will be able to delete a catch all account for the domain without notice. This is pretty much what I did for now... I think the best is to fix this issue making possible that the default alias is inserted in the valias table like the others (if you use aliases in database), or to let the function check both file and database. Exactly. It needs to check for a .qmail file first since that is how qmail itself handles the situation, and I don't want to mess with that. For all aliases, valias needs to check for a file first and if it is found manage the alias in the file. If there is no existing file then it can go to the database. The work actually needs to be done in valias, but I don't want to do anything there until I have a nice stable release for people to stick with while I add some new features. Rick
Re: [vchkpw] Decided to make the plunge today on late version of vpopmail
Quey wrote: Rick on a kind of related note, did you see one of my posts in the thread: adduser processing times CDB from 27th of the 7th ? Looking at that thread... slow loading of users: I have added a note in the TODO to add an -O option to vadduser like the option in vadddomain, so it can only update the .cdb file once at the end of a batch of adds. Right now I'm pretty sure the .cdb is updated for each one. Missing some files in vdeldomain: The current code uses unlink(), I don't think I want to do anything more when deleting files. Considering that you had to e2fsck the disk I'm not sure I could. I don't believe vpopmail ever creates files or directories with a ? in the name and have no idea where they may have come from. Learn-Password: learn-password can be turned on at any time with /configure; make; make install. Be sure to re-compile everything that uses vpopmail Rick
[vchkpw] vdeluser isCatchall mysql segfault (was Re: [vchkpw] Decided to make the plunge today on late version of vpopmail)
Hi Steve,
i think the problem is related to something i posted some days ago but
without any answer.
Look at the post :
Re: [vchkpw] vdeluser segfault posted on 19/08/2007 be my answering
to a previous post.
Let me know if it is the case.
Bye,
Marcello
Steve Cole wrote:
On two (low volume) machines with vpopmail interfaced with mysql as the
data
store, it seemed to work successfully. I updated the database schema
with no
issues. This is with v5.4.21
However, when I do a vdeluser on either machine, the program
segfaults. On
both systems.
I tried various things like making sure my LDFLAGS= and CFLAGS=-O
only to
be sure it wasn't a compiler issue, without luck.
So, to be clear, this is on two Debian Etch machines with GCC 4.1.1-15
and
MySQL 5.0.32-7etch1 installed.
Going back to 5.4.17 seemed to be no issue at all and of course, it
works as
expected.
I would concur. From a cursory look at the logic, I don't see how this
really works when using MySQL aliases. I commented out the entire catchall
check when moving to 5.4.19 simply to keep things working.
Regards,
-jc
--- vpopmail.c.orig 2007-08-23 16:46:31.0 -0700
+++ vpopmail.c 2007-08-23 17:03:51.0 -0700
@@ -1733,9 +1733,9 @@
}
/* Make sure we are not the email address of the catchall account */
- if ( isCatchall( user, domain, Dir )) {
-return(VA_CANNOT_DELETE_CATCHALL);
-}
+ // if ( isCatchall( user, domain, Dir )) {
+ // return(VA_CANNOT_DELETE_CATCHALL);
+ // }
#ifdef ONCHANGE_SCRIPT
/* tell other programs that data has changed */
Re: [vchkpw] Decided to make the plunge today on late version of vpopmail
Hi Steve, i think the problem is related to something i posted some days ago but without any answer. Look at the post : Re: [vchkpw] vdeluser segfault posted on 19/08/2007 be my answering to a previous post. Let me know if it is the case. Bye, Marcello Steve Cole wrote: On two (low volume) machines with vpopmail interfaced with mysql as the data store, it seemed to work successfully. I updated the database schema with no issues. This is with v5.4.21 However, when I do a vdeluser on either machine, the program segfaults. On both systems. I tried various things like making sure my LDFLAGS= and CFLAGS=-O only to be sure it wasn't a compiler issue, without luck. So, to be clear, this is on two Debian Etch machines with GCC 4.1.1-15 and MySQL 5.0.32-7etch1 installed. Going back to 5.4.17 seemed to be no issue at all and of course, it works as expected.
Re: [vchkpw] vdeluser isCatchall mysql segfault (was Re: [vchkpw] Decided to make the plunge today on late version of vpopmail)
Can someone who can see it fail please send me your ./configure options
so maybe I can make it fail here?
Rick
Marcello Lupo wrote:
Hi,
i agree with your method to bybass the problem but in this way you will
be able to delete a catch all account
for the domain without notice.
I think the best is to fix this issue making possible that the default
alias is inserted in the valias table like the others
(if you use aliases in database), or to let the function check both file
and database.
Moreover if it don't found the default alias in the database (this is
what happen now) it should segfault but assume
that the account is not catchall.
I'm not a coder and i'm not able to do it.
Bye,
Marcello
Japheth Cleaver wrote:
Hi Steve,
i think the problem is related to something i posted some days ago but
without any answer.
Look at the post :
Re: [vchkpw] vdeluser segfault posted on 19/08/2007 be my answering
to a previous post.
Let me know if it is the case.
Bye,
Marcello
Steve Cole wrote:
On two (low volume) machines with vpopmail interfaced with mysql as the
data
store, it seemed to work successfully. I updated the database schema
with no
issues. This is with v5.4.21
However, when I do a vdeluser on either machine, the program
segfaults. On
both systems.
I tried various things like making sure my LDFLAGS= and CFLAGS=-O
only to
be sure it wasn't a compiler issue, without luck.
So, to be clear, this is on two Debian Etch machines with GCC 4.1.1-15
and
MySQL 5.0.32-7etch1 installed.
Going back to 5.4.17 seemed to be no issue at all and of course, it
works as
expected.
I would concur. From a cursory look at the logic, I don't see how this
really works when using MySQL aliases. I commented out the entire
catchall
check when moving to 5.4.19 simply to keep things working.
Regards,
-jc
--- vpopmail.c.orig 2007-08-23 16:46:31.0 -0700
+++ vpopmail.c 2007-08-23 17:03:51.0 -0700
@@ -1733,9 +1733,9 @@
}
/* Make sure we are not the email address of the catchall account */
- if ( isCatchall( user, domain, Dir )) {
-return(VA_CANNOT_DELETE_CATCHALL);
-}
+ // if ( isCatchall( user, domain, Dir )) {
+ // return(VA_CANNOT_DELETE_CATCHALL);
+ // }
#ifdef ONCHANGE_SCRIPT
/* tell other programs that data has changed */
Re: [vchkpw] Decided to make the plunge today on late version of vpopmail
Rick Widmer wrote: Quey wrote: Rick on a kind of related note, did you see one of my posts in the thread: adduser processing times CDB from 27th of the 7th ? Looking at that thread... slow loading of users: I have added a note in the TODO to add an -O option to vadduser like the option in vadddomain, so it can only update the .cdb file once at the end of a batch of adds. Right now I'm pretty sure the .cdb is updated for each one. That would be good, however we have since moved to mysql and it flies, but it would still be a good idea ofr those still using cdb for sure. Missing some files in vdeldomain: The current code uses unlink(), I don't think I want to do anything more when deleting files. Considering that you had to e2fsck the disk I'm not sure I could. I don't believe vpopmail ever creates files or directories with a ? in the name and have no idea where they may have come from. They were whats left after vdeldomain, they were directory names and stuff like what was there before named correctly, after i deleted the domain something must have had a hissy somewhere and was unable to remove them so renamed them with a preceeding ?, that must have something to do with unlinks work I guess, when I next have to build another box I'll do a dummy run and delete it and see what happens, maybe it was coincidence and was purely a one-off case, I wouldnt suggest doing it yourself on anything production :) Cheers Q
Re: [vchkpw] vdeluser isCatchall mysql segfault (was Re: [vchkpw] Decided to make the plunge today on late version of vpopmail)
Hi,
i agree with your method to bybass the problem but in this way you will
be able to delete a catch all account
for the domain without notice.
I think the best is to fix this issue making possible that the default
alias is inserted in the valias table like the others
(if you use aliases in database), or to let the function check both file
and database.
Moreover if it don't found the default alias in the database (this is
what happen now) it should segfault but assume
that the account is not catchall.
I'm not a coder and i'm not able to do it.
Bye,
Marcello
Japheth Cleaver wrote:
Hi Steve,
i think the problem is related to something i posted some days ago but
without any answer.
Look at the post :
Re: [vchkpw] vdeluser segfault posted on 19/08/2007 be my answering
to a previous post.
Let me know if it is the case.
Bye,
Marcello
Steve Cole wrote:
On two (low volume) machines with vpopmail interfaced with mysql as the
data
store, it seemed to work successfully. I updated the database schema
with no
issues. This is with v5.4.21
However, when I do a vdeluser on either machine, the program
segfaults. On
both systems.
I tried various things like making sure my LDFLAGS= and CFLAGS=-O
only to
be sure it wasn't a compiler issue, without luck.
So, to be clear, this is on two Debian Etch machines with GCC 4.1.1-15
and
MySQL 5.0.32-7etch1 installed.
Going back to 5.4.17 seemed to be no issue at all and of course, it
works as
expected.
I would concur. From a cursory look at the logic, I don't see how this
really works when using MySQL aliases. I commented out the entire catchall
check when moving to 5.4.19 simply to keep things working.
Regards,
-jc
--- vpopmail.c.orig 2007-08-23 16:46:31.0 -0700
+++ vpopmail.c 2007-08-23 17:03:51.0 -0700
@@ -1733,9 +1733,9 @@
}
/* Make sure we are not the email address of the catchall account */
- if ( isCatchall( user, domain, Dir )) {
-return(VA_CANNOT_DELETE_CATCHALL);
-}
+ // if ( isCatchall( user, domain, Dir )) {
+ // return(VA_CANNOT_DELETE_CATCHALL);
+ // }
#ifdef ONCHANGE_SCRIPT
/* tell other programs that data has changed */
Re: [vchkpw] vdeluser isCatchall mysql segfault (was Re: [vchkpw] Decided to make the plunge today on late version of vpopmail)
Hi Rick,
this is my configure options with vpopmail-5.4.19 :
./configure --enable-roaming-users
--enable-tcpserver-file=/home/vpopmail/etc/tcp.smtp
--enable-relay-clear-minutes=15 \
--enable-onchange-script --enable-auth-module=mysql --enable-logging=p
--enable-mysql-limits --enable-valias \
--enable-incdir=/usr/include/mysql --enable-libdir=/usr/lib64/mysql
--disable-clear-passwd
It is making the problem on this installation.
Thanks,
Bye,
Marcello
Rick Widmer wrote:
Can someone who can see it fail please send me your ./configure
options so maybe I can make it fail here?
Rick
Marcello Lupo wrote:
Hi,
i agree with your method to bybass the problem but in this way you
will be able to delete a catch all account
for the domain without notice.
I think the best is to fix this issue making possible that the
default alias is inserted in the valias table like the others
(if you use aliases in database), or to let the function check both
file and database.
Moreover if it don't found the default alias in the database (this is
what happen now) it should segfault but assume
that the account is not catchall.
I'm not a coder and i'm not able to do it.
Bye,
Marcello
Japheth Cleaver wrote:
Hi Steve,
i think the problem is related to something i posted some days ago but
without any answer.
Look at the post :
Re: [vchkpw] vdeluser segfault posted on 19/08/2007 be my
answering
to a previous post.
Let me know if it is the case.
Bye,
Marcello
Steve Cole wrote:
On two (low volume) machines with vpopmail interfaced with mysql
as the
data
store, it seemed to work successfully. I updated the database schema
with no
issues. This is with v5.4.21
However, when I do a vdeluser on either machine, the program
segfaults. On
both systems.
I tried various things like making sure my LDFLAGS= and CFLAGS=-O
only to
be sure it wasn't a compiler issue, without luck.
So, to be clear, this is on two Debian Etch machines with GCC
4.1.1-15
and
MySQL 5.0.32-7etch1 installed.
Going back to 5.4.17 seemed to be no issue at all and of course, it
works as
expected.
I would concur. From a cursory look at the logic, I don't see how this
really works when using MySQL aliases. I commented out the entire
catchall
check when moving to 5.4.19 simply to keep things working.
Regards,
-jc
--- vpopmail.c.orig 2007-08-23 16:46:31.0 -0700
+++ vpopmail.c 2007-08-23 17:03:51.0 -0700
@@ -1733,9 +1733,9 @@
}
/* Make sure we are not the email address of the catchall account */
- if ( isCatchall( user, domain, Dir )) {
-return(VA_CANNOT_DELETE_CATCHALL);
-}
+ // if ( isCatchall( user, domain, Dir )) {
+ // return(VA_CANNOT_DELETE_CATCHALL);
+ // }
#ifdef ONCHANGE_SCRIPT
/* tell other programs that data has changed */
Re: [vchkpw] vdeluser isCatchall mysql segfault (was Re: [vchkpw] Decided to make the plunge today on late version of vpopmail)
Marcello Lupo wrote: Hi Rick, this is my configure options with vpopmail-5.4.19 : ./configure --enable-roaming-users --enable-tcpserver-file=/home/vpopmail/etc/tcp.smtp --enable-relay-clear-minutes=15 \ --enable-onchange-script --enable-auth-module=mysql --enable-logging=p --enable-mysql-limits --enable-valias \ --enable-incdir=/usr/include/mysql --enable-libdir=/usr/lib64/mysql --disable-clear-passwd Cool! It failed. Now on to a fix...
Re: [vchkpw] Decided to make the plunge today on late version of vpopmail
On two (low volume) machines with vpopmail interfaced with mysql as the data store, it seemed to work successfully. I updated the database schema with no issues. This is with v5.4.21 Did you follow the UPGRADE document in 5.4.21 exactly and rename the 'domain' columns to 'pw_domain'? If so, then your DB is now wrong. Apparently the fix I posted about the UPGRADE document didn't make it into 5.4.21... The vpopmail table, and ONLY the vpopmail table, has a column named 'pw_domain'. Every other table should have a 'domain' column, as the earlier UPGRADE document listed (which incorrectly listed vpopmail as having a 'domain' column). The other suggestion I can make is try stracing the new version and see where it fails... Josh Joshua Megerman SJGames MIB #5273 - OGRE AI Testing Division You can't win; You can't break even; You can't even quit the game. - Layman's translation of the Laws of Thermodynamics [EMAIL PROTECTED]
Re: [vchkpw] Decided to make the plunge today on late version of vpopmail
On Thursday 06 September 2007, Joshua Megerman wrote: The vpopmail table, and ONLY the vpopmail table, has a column named 'pw_domain'. Every other table should have a 'domain' column, as the earlier UPGRADE document listed (which incorrectly listed vpopmail as having a 'domain' column). No, thankfully. I was the one who reported the inconsistency in the upgrade document when *.20 rolled. I should have mentioned that this same thing happened to one of my test boxes when I tried *.20 but it didn't on another (which it turns out is CDB because it handles two personal domains). So *.20 likely has the problem as well, I didn't delve that deeply the first time. I straced the command and it opens a MySQL connection, does a MySQL query and then segfaults immediately. As I said, the behaviour doesn't occur in the CDB binary. -- -- Cheers, Steve
Re: [vchkpw] Decided to make the plunge today on late version of vpopmail
Rick Widmer wrote: Steve Cole wrote: On two (low volume) machines with vpopmail interfaced with mysql as the data store, it seemed to work successfully. I updated the database schema with no issues. This is with v5.4.21 However, when I do a vdeluser on either machine, the program segfaults. On both systems. I tried it here with --enable-auth-module=mysql but don't get the segfault. So can you send me: The vpopmail ./configure options you are using The actual user/domain you are trying to delete The strace output You can send it private if you wish... Going back to 5.4.17 seemed to be no issue at all and of course, it works as expected. Nothing within vdeluser changed between those versions, but there were changes in vpopmail.c, but nothing that stands out. Rick Widmer wrote: Steve Cole wrote: On two (low volume) machines with vpopmail interfaced with mysql as the data store, it seemed to work successfully. I updated the database schema with no issues. This is with v5.4.21 However, when I do a vdeluser on either machine, the program segfaults. On both systems. I tried it here with --enable-auth-module=mysql but don't get the segfault. So can you send me: The vpopmail ./configure options you are using The actual user/domain you are trying to delete The strace output You can send it private if you wish... Going back to 5.4.17 seemed to be no issue at all and of course, it works as expected. Nothing within vdeluser changed between those versions, but there were changes in vpopmail.c, but nothing that stands out. Rick on a kind of related note, did you see one of my posts in the thread: adduser processing times CDB from 27th of the 7th ?
Re: [vchkpw] Decided to make the plunge today on late version of vpopmail
Steve Cole wrote: On two (low volume) machines with vpopmail interfaced with mysql as the data store, it seemed to work successfully. I updated the database schema with no issues. This is with v5.4.21 However, when I do a vdeluser on either machine, the program segfaults. On both systems. I tried it here with --enable-auth-module=mysql but don't get the segfault. So can you send me: The vpopmail ./configure options you are using The actual user/domain you are trying to delete The strace output You can send it private if you wish... Going back to 5.4.17 seemed to be no issue at all and of course, it works as expected. Nothing within vdeluser changed between those versions, but there were changes in vpopmail.c, but nothing that stands out.
