Обсуждение: problem with pgadmin 1.8.2 to keep connection to postgresql 8.2.6

Hi,

i installed the pgadmin3 rpm for opensuse 10.2 i586, from the pgadmin 
download page. I can connect to the database, and also run some few queries, 
but then, sometimes after the second, sometimes a bit later, maybe after the 
fifth, or tenth time, the query does not succeed anymore and in the result 
field in pgadmin3 it says ********error*********
When I then try again to submit the query, a small empty window pops up, and 
then pgadmin3 hangs. I either have to kill it, or it dies by itself.
When pgadmin it dies, I see the following on the console:

(pgadmin3:5689): Gdk-CRITICAL **: gdk_window_set_geometry_hints: assertion 
`GDK_IS_WINDOW (window)' failed

(pgadmin3:5689): Gdk-CRITICAL **: gdk_window_resize: assertion 
`GDK_IS_WINDOW (window)' failed
Xlib: unexpected async reply (sequence 0xa3a60)!
Xlib: sequence lost (0xb0000 > 0xa3a61) in reply type 0x0!

I also tried to run pgadmin3 --sync, as on another time when it crashed, it 
suggested to use this parameter. But then, the problem was not reproducible, 
because the query seems to take forever, the counter of miliseconds in the 
lower right corner was at about 50000 when I stopped it. The query run from 
pgadmin without --sync parameter took only about 40ms.

The query that I submit is very easy: select * from segment; in that table, 
there are only twenty rows in it, and I can do the same statement from 
commandline via psql without any problems.

The database is running on a SLES10SP1 x86_64, the client is a opensuse 
10.2, i586.

Has anybody seen this, and an idea, how to fix it?

kind regards
Sebastian

On Fri, May 16, 2008 at 12:05 PM, Sebastian Reitenbach
<sebastia@l00-bugdead-prods.de> wrote:
> I also tried to run pgadmin3 --sync, as on another time when it crashed, it
> suggested to use this parameter. But then, the problem was not reproducible,
> because the query seems to take forever, the counter of miliseconds in the
> lower right corner was at about 50000 when I stopped it. The query run from
> pgadmin without --sync parameter took only about 40ms.

Well that's, umm surprising. There is no --sync option in pgAdmin - in
fact I get an error if I try to use it on Windows or Mac (I don't have
a GTK system here atm). Can you get an exact copy of the text which
recommended you use it?


-- 
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

"Dave Page" <dpage@pgadmin.org> wrote: 
> On Fri, May 16, 2008 at 12:05 PM, Sebastian Reitenbach
> <sebastia@l00-bugdead-prods.de> wrote:
> > I also tried to run pgadmin3 --sync, as on another time when it crashed, 
it
> > suggested to use this parameter. But then, the problem was not 
reproducible,
> > because the query seems to take forever, the counter of miliseconds in 
the
> > lower right corner was at about 50000 when I stopped it. The query run 
from
> > pgadmin without --sync parameter took only about 40ms.
> 
> Well that's, umm surprising. There is no --sync option in pgAdmin - in
> fact I get an error if I try to use it on Windows or Mac (I don't have
> a GTK system here atm). Can you get an exact copy of the text which
> recommended you use it?
Afaik the error message came from the X server, so yes, not a pgadmin3 
parameter.
It said sth. like:
When an X application crashes, the error message is asynchron, and therefore 
usually not shows the real cause of the problem. When you want to run the 
application from within gdb, then use --sync, to get the right backtrace.
The --sync is I think automatically added to each X application, from the X 
server, or window manager, whoever there is responsible.

When I get the error again, then I cut 'n past the output in here, I got 
this only once.

Sebastian

On Fri, May 16, 2008 at 4:16 PM, Sebastian Reitenbach
<sebastia@l00-bugdead-prods.de> wrote:
> "Dave Page" <dpage@pgadmin.org> wrote:
>> On Fri, May 16, 2008 at 12:05 PM, Sebastian Reitenbach
>> <sebastia@l00-bugdead-prods.de> wrote:
>> > I also tried to run pgadmin3 --sync, as on another time when it crashed,
> it
>> > suggested to use this parameter. But then, the problem was not
> reproducible,
>> > because the query seems to take forever, the counter of miliseconds in
> the
>> > lower right corner was at about 50000 when I stopped it. The query run
> from
>> > pgadmin without --sync parameter took only about 40ms.
>>
>> Well that's, umm surprising. There is no --sync option in pgAdmin - in
>> fact I get an error if I try to use it on Windows or Mac (I don't have
>> a GTK system here atm). Can you get an exact copy of the text which
>> recommended you use it?
> Afaik the error message came from the X server, so yes, not a pgadmin3
> parameter.
> It said sth. like:
> When an X application crashes, the error message is asynchron, and therefore
> usually not shows the real cause of the problem. When you want to run the
> application from within gdb, then use --sync, to get the right backtrace.
> The --sync is I think automatically added to each X application, from the X
> server, or window manager, whoever there is responsible.

Hmm, OK - I did wonder if it were something like that.

> When I get the error again, then I cut 'n past the output in here, I got
> this only once.

Please do. FWIW, there are no known bugs in pgAdmin that would cause
connections to drop as you describe (in fact, the connections are
handled by libpq, so it would be largely out of our control anyway) -
and it should be handled gracefully by pgAdmin when it does happen for
other reasons. I have heard a report in the past that for one person
that wasn't the case, but I was unable to reproduce any problems when
deliberately causing connection problems.

-- 
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Hi,
"Dave Page" <dpage@pgadmin.org> wrote: 
> On Fri, May 16, 2008 at 4:16 PM, Sebastian Reitenbach
> <sebastia@l00-bugdead-prods.de> wrote:
> > "Dave Page" <dpage@pgadmin.org> wrote:
> >> On Fri, May 16, 2008 at 12:05 PM, Sebastian Reitenbach
> >> <sebastia@l00-bugdead-prods.de> wrote:
> >> > I also tried to run pgadmin3 --sync, as on another time when it 
crashed,
> > it
> >> > suggested to use this parameter. But then, the problem was not
> > reproducible,
> >> > because the query seems to take forever, the counter of miliseconds 
in
> > the
> >> > lower right corner was at about 50000 when I stopped it. The query 
run
> > from
> >> > pgadmin without --sync parameter took only about 40ms.
> >>
> >> Well that's, umm surprising. There is no --sync option in pgAdmin - in
> >> fact I get an error if I try to use it on Windows or Mac (I don't have
> >> a GTK system here atm). Can you get an exact copy of the text which
> >> recommended you use it?
> > Afaik the error message came from the X server, so yes, not a pgadmin3
> > parameter.
> > It said sth. like:
> > When an X application crashes, the error message is asynchron, and 
therefore
> > usually not shows the real cause of the problem. When you want to run 
the
> > application from within gdb, then use --sync, to get the right 
backtrace.
> > The --sync is I think automatically added to each X application, from 
the X
> > server, or window manager, whoever there is responsible.
> 
> Hmm, OK - I did wonder if it were something like that.
> 
> > When I get the error again, then I cut 'n past the output in here, I got
> > this only once.
> 
> Please do. FWIW, there are no known bugs in pgAdmin that would cause
> connections to drop as you describe (in fact, the connections are
> handled by libpq, so it would be largely out of our control anyway) -
> and it should be handled gracefully by pgAdmin when it does happen for
> other reasons. I have heard a report in the past that for one person
> that wasn't the case, but I was unable to reproduce any problems when
> deliberately causing connection problems.

sorry, it took me a while to get back to the machines again, however, here 
is the error output cut 'n pasted:

So, after submitting the same query, for about 10 times, then instead of the 
result set, the following was shown in the sql output window of pgadmin3:
********** Error **********

Then, when I try to submit the query again, an error popup window shows up, 
stating the following:

==================================
An error has occurred:

no connection to the server
==================================
and at the same time, on the console the following appears:

==================================
Xlib: sequence lost (0xd5bf8 > 0xc5bfa) in reply type 0xf!
Xlib: unexpected async reply (sequence 0xc5c81)!
==================================

When I then try to close the error window, it is not closing, I klick 
the "close" or "ok" or however the button is called, the button several 
times, and then pgadmin3 disappears at all, leaving the following message on 
the console:

==================================
The program 'pgadmin3' received an X Window System error.
This probably reflects a bug in the program.
The error was 'BadAtom (invalid Atom parameter)'. (Details: serial 967950 error_code 5 request_code 18 minor_code 0)
(Noteto programmers: normally, X errors are reported asynchronously;  that is, you will receive the error a while after
causingit.  To debug your program, run it with the --sync command line  option to change this behavior. You can then
geta meaningful  backtrace from your debugger if you break on the gdk_x_error() function.)
 
==================================

I also retested, with psql from commandline, using the same query,
and I was able to issue it about 50 times in a fast row, before I stopped 
again.

I also tried pgadmin 1.8.1, and I did ran into the same problem, but it 
seems that the problem does not occur so fast, but maybe that was just 
coincidence.

Sebastian

On Tue, May 20, 2008 at 9:01 AM, Sebastian Reitenbach
<sebastia@l00-bugdead-prods.de> wrote:

> I also retested, with psql from commandline, using the same query,
> and I was able to issue it about 50 times in a fast row, before I stopped
> again.
>
> I also tried pgadmin 1.8.1, and I did ran into the same problem, but it
> seems that the problem does not occur so fast, but maybe that was just
> coincidence.

[sorry for the delay - I was travelling last week]

I've been unable to reproduce this at all. Are your client and server
one different hosts, or even different networks? Anything that might
explain why it's disconnecting in the first place? What's odd is that
psql works, as both psql and pgAdmin use libpq to handle the
connection so there should be no difference (unless perhaps you are
using a Unix socket with psql, and the network is a little flakey). Is
there anything in the server log?

-- 
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com