 |
 |
Hi Tommaso,
OK, this is now not critical as I seem to have found the problem. However, a backtrace would still be good to have as assurance. Yes, it is a bug when invalid login data is encountered. Now we have tto find out why this actually happened.
What is wngw.ifca.es actually running (i.e. what cmsd version).
Andy
On Mon, 24 Aug 2015, Tommaso Boccali wrote:
ciao andrew, tomorrow I can try but the fact is taht today in the end I had
to downgrade, since it is a production server.
so i have to reupgrade, take the snapshot and go back as fast as possible :(
ciao ciao
tom
On Mon, Aug 24, 2015 at 9:47 PM, Andrew Hanushevsky <[log in to unmask]>
wrote:
Hi Tommaso,
Both daemons (xrootd and cmsd) will exit if you attempt to run them as
root. This is a security feature. You can run them as root but only after
specifically confirming this via command line options (i.e. you accept the
risks). As for the SEGV, that's clearly a bug. Is it possible to get a
stack trace of the thread that got the SEGV? Please make sure to install
the debug RPM so we can get actual line numbers.
Andy
On Mon, 24 Aug 2015, Tommaso Boccali wrote:
uhm,
- the strace line was just my fault, I was trying running as root on the
command line
- when I retried with user xrootd, I get instead the lines below, which
terminate with a segv (*)
so the last message is consistent with the one in the logs:
[pid 65395] writev(2, [{"150824 10:09:53 65395 ", 22}, {"Pup", 3}, {": ",
2}, {"buffer overrun unpacking", 24}, {" ", 1}, {"short arg 0: ident.",
19}, {"\n", 1}], 7) = 72
[pid 65395] gettid() = 65395
[pid 65395] writev(2, [{"150824 10:09:53 65395 ", 22}, {"Login", 5}, {":
",
2}, {"wngw.ifca.es", 12}, {" ", 1}, {"login failed;", 13}, {" ", 1},
{"invalid login data", 18}, {"\n", 1}], 9) = 75
*:
[pid 65395] <... gettid resumed> ) = 65395
[pid 65395] write(2, "150824 10:09:53 65395 ", 22) = 22
[pid 65395] write(2, "Xrd", 3) = 3
[pid 65395] write(2, "Inet", 4) = 4
[pid 65395] write(2, ": ", 2) = 2
[pid 65395] write(2, "Accepted connection from ", 25) = 25
[pid 65395] write(2, "23", 2) = 2
[pid 65395] write(2, "@", 1) = 1
[pid 65395] write(2, "wngw.ifca.es", 12) = 12
[pid 65395] write(2, "\n", 1) = 1
[pid 65395] futex(0x6473c8, FUTEX_WAKE_PRIVATE, 1) = 0
[pid 65395] poll([{fd=23, events=POLLIN|POLLRDNORM}], 1, 1000) = 1
([{fd=23, revents=POLLIN|POLLRDNORM}])
[pid 65395] recvfrom(23, "\0\0\0\0\0\0\0\0", 8, MSG_PEEK, NULL, NULL) = 8
[pid 65395] gettid() = 65395
[pid 65395] write(2, "150824 10:09:53 65395 ", 22) = 22
[pid 65395] write(2, "Xrd", 3) = 3
[pid 65395] write(2, "Protocol", 8) = 8
[pid 65395] write(2, ": ", 2) = 2
[pid 65395] write(2, "matched protocol ", 17) = 17
[pid 65395] write(2, "cmsd", 4) = 4
[pid 65395] write(2, "\n", 1) = 1
[pid 65395] epoll_ctl(12, EPOLL_CTL_ADD, 23, {0, {u32=4160757352,
u64=140071634214504}}) = 0
[pid 65395] gettid() = 65395
[pid 65395] write(2, "150824 10:09:53 65395 ", 22) = 22
[pid 65395] write(2, "?:[log in to unmask]", 17) = 17
[pid 65395] write(2, " ", 1) = 1
[pid 65395] write(2, "Xrd", 3) = 3
[pid 65395] write(2, "Poll", 4) = 4
[pid 65395] write(2, ": ", 2) = 2
[pid 65395] write(2, "FD ", 3) = 3
[pid 65395] write(2, "23", 2) = 2
[pid 65395] write(2, " attached to poller ", 20) = 20
[pid 65395] write(2, "2", 1) = 1
[pid 65395] write(2, "; num=", 6) = 6
[pid 65395] write(2, "1", 1) = 1
[pid 65395] write(2, "\n", 1) = 1
[pid 65395] poll([{fd=23, events=POLLIN|POLLRDNORM}], 1, 5000) = 1
([{fd=23, revents=POLLIN|POLLRDNORM}])
[pid 65395] recvfrom(23, "\0\0\0\0\0\0\0\0", 8, 0, NULL, NULL) = 8
[pid 65395] gettid() = 65395
[pid 65395] writev(2, [{"150824 10:09:53 65395 ", 22}, {"Pup", 3}, {": ",
2}, {"buffer overrun unpacking", 24}, {" ", 1}, {"short arg 0: ident.",
19}, {"\n", 1}], 7) = 72
[pid 65395] gettid() = 65395
[pid 65395] writev(2, [{"150824 10:09:53 65395 ", 22}, {"Login", 5}, {":
",
2}, {"wngw.ifca.es", 12}, {" ", 1}, {"login failed;", 13}, {" ", 1},
{"invalid login data", 18}, {"\n", 1}], 9) = 75
[pid 65395] --- SIGSEGV (Segmentation fault) @ 0 (0) ---
Process 65395 detached
[pid 65394] +++ killed by SIGSEGV (core dumped) +++
[pid 65389] +++ killed by SIGSEGV (core dumped) +++
[pid 65397] +++ killed by SIGSEGV (core dumped) +++
[pid 65396] +++ killed by SIGSEGV (core dumped) +++
[pid 65379] +++ killed by SIGSEGV (core dumped) +++
[pid 65383] +++ killed by SIGSEGV (core dumped) +++
[pid 65391] +++ killed by SIGSEGV (core dumped) +++
[pid 65392] +++ killed by SIGSEGV (core dumped) +++
[pid 65393] +++ killed by SIGSEGV (core dumped) +++
[pid 65390] +++ killed by SIGSEGV (core dumped) +++
[pid 65386] +++ killed by SIGSEGV (core dumped) +++
[pid 65385] +++ killed by SIGSEGV (core dumped) +++
[pid 65388] +++ killed by SIGSEGV (core dumped) +++
[pid 65387] +++ killed by SIGSEGV (core dumped) +++
[pid 65384] +++ killed by SIGSEGV (core dumped) +++
[pid 65382] +++ killed by SIGSEGV (core dumped) +++
[pid 65381] +++ killed by SIGSEGV (core dumped) +++
[pid 65380] +++ killed by SIGSEGV (core dumped) +++
[pid 65378] +++ killed by SIGSEGV (core dumped) +++
[pid 65377] +++ killed by SIGSEGV (core dumped) +++
[pid 65376] +++ killed by SIGSEGV (core dumped) +++
[pid 65375] +++ killed by SIGSEGV (core dumped) +++
+++ killed by SIGSEGV (core dumped) +++
On Mon, Aug 24, 2015 at 10:02 AM, Jan Iven <[log in to unmask]> wrote:
On 08/24/2015 09:51 AM, Tommaso Boccali wrote:
Ciao, I am trying to upgrade one of the CMS EU redirectors from 3.3.6 to
4.2.2 (with no configuration changed at first approx)Well, that line ought to end up either on STDERR somewhere or in some log
The problem is that the main cmsd seems to die soon after start. No real
message in the logs, but with strace I see a very suspect
writev(2, [{"Copr. 2007 Stanford University/"..., 42}, {"\n", 1}], 2) =
43
geteuid() = 0
gettid() = 53572
writev(2, [{"150824 09:47:27 53572 ", 22}, {"Config", 6}, {": ", 2},
{"Security reasons prohibit cmsd r"..., 73}, {"\n", 1}], 5) = 104
file. Alternatively, suggest "strace -s 1024" to get the full error
message..
Cheers
jan
--
Tommaso Boccali
INFN Pisa
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
--
Tommaso Boccali
INFN Pisa
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1