13 Replies Latest reply on Aug 16, 2010 8:02 AM by 791601

    SQL*Plus 10.2.0.1.0 does not work on RHEL3

    user458167
      I have a server running Red Hat Enterprise Linux 3 ES with Update 6, which has not been rebooted for 288 days (it's still running the kernel version 2.4.21-27.ELsmp, from Update 4). I need to run SQL*Plus on this server. The rpm packages for instant client 10.1.0.3 and 10.1.0.4 allow me to run SQL*Plus without any problems. However, when I install the instant client packages version 10.2.0.1 (both zip and rpm do the same), SQL*Plus just hangs when I try to start it. When I change the versions of SQL*Plus, I always edit /etc/ld.so.conf to point to the right directory with libraries and then run ldconfig. When I run sqlplus 10.2.0.1 through strace, I see that it ends up with this:

      gettimeofday({1129717666, 622797}, NULL) = 0
      access("/etc/sqlnet.ora", F_OK) = -1 ENOENT (No such file or directory)
      access("./network/admin/sqlnet.ora", F_OK) = -1 ENOENT (No such file or directory)
      access("/etc/sqlnet.ora", F_OK) = -1 ENOENT (No such file or directory)
      access("./network/admin/sqlnet.ora", F_OK) = -1 ENOENT (No such file or directory)
      fcntl64(155815832, F_SETFD, FD_CLOEXEC) = -1 EBADF (Bad file descriptor)
      times(NULL) = -1808543702
      times(NULL) = -1808543702
      times(NULL) = -1808543702
      times(NULL) = -1808543702
      times(NULL) = -1808543702
      times(NULL) = -1808543702
      times(NULL) = -1808543702

      It tries to look up the same files twice - strange, but not necessarrily a problem. However, the times(NULL) call repeating untill I stop the process definitely is a problem. The strace output from SQL*Plus 10.1.0.4.0 shows that times(NULL) is called just once there. Is this a bug in SQL*Plus 10.2.0.1.0?
        • 1. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
          Laurenz Albe
          I don't know what's wrong, but you could run a trace:

          add the following to ./network/admin/sqlnet.ora:

          TRACE_DIRECTORY_CLIENT= /path/to/dir/for/logfile
          TRACE_LEVEL_CLIENT= SUPPORT

          Then try again. The resulting trace file is large, but may give you a clue.

          Yours,
          Laurenz Albe
          • 2. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
            user458167
            Thanks for the hint. I have created a new file ./network/admin/sqlnet.ora with the content you suggested. However, it did not help. Nothing appeared in the specified directory - the process probably hangs before it would start creating any data there.

            This is the new output from strace:

            gettimeofday({1131473447, 389346}, NULL) = 0
            access("./network/admin/sqlnet.ora", F_OK) = 0
            open("./network/admin/sqlnet.ora", O_RDONLY|O_LARGEFILE) = 3
            fcntl64(3, F_SETFD, FD_CLOEXEC) = 0
            fstat64(3, {st_mode=S_IFREG|0664, st_size=66, ...}) = 0
            mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb755d000
            read(3, "TRACE_DIRECTORY_CLIENT=/mnt/big0"..., 4096) = 66
            read(3, "", 4096) = 0
            close(3) = 0
            munmap(0xb755d000, 4096) = 0
            access("./network/admin/sqlnet.ora", F_OK) = 0
            open("./network/admin/sqlnet.ora", O_RDONLY|O_LARGEFILE) = 3
            fcntl64(3, F_SETFD, FD_CLOEXEC) = 0
            fstat64(3, {st_mode=S_IFREG|0664, st_size=66, ...}) = 0
            mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb755d000
            read(3, "TRACE_DIRECTORY_CLIENT=/mnt/big0"..., 4096) = 66
            read(3, "", 4096) = 0
            close(3) = 0
            munmap(0xb755d000, 4096) = 0
            fcntl64(156324384, F_SETFD, FD_CLOEXEC) = -1 EBADF (Bad file descriptor)
            times(NULL) = -1632972122
            times(NULL) = -1632972122
            times(NULL) = -1632972122
            times(NULL) = -1632972122
            times(NULL) = -1632972122

            Again, the times(NULL) calls repeat untill I kill the process. Additionally, I do not understand why it does not look for /etc/sqlnet.ora anymore.
            • 3. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
              466776
              I had the same issue with sqlplus hung on times(NULL). If your uptime is greater than 248 days, you may want to reboot and try again as times(NULL) is possibly overflowing.
              • 4. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                user458167
                I thought that the long uptime might be an issue. However, considering that 10.1.0.3 and 10.1.0.4 work fine, I hope that it would not be a big problem to persuade somebody from Oracle that this is a bug in 10.2 and that it should be fixed. Unfortunatelly, even though our company has bought many licences for various Oracle products and I have access to metalink, I don't know what is the proper process for reporting bugs in instant client (which is a free product, if I understand it correctly).
                • 5. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                  Laurenz Albe
                  My box has an uptime of 141 days.
                  As soon as I hit the bug, I'll open a TAR.
                  • 6. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                    352290
                    To report a bug with Instant Client, please log on to Metalink, and detail the problem in a TAR. I hope that helps!
                    • 7. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                      user458167
                      OK, I have filed TAR 4867480.993 at metalink. I did not find any option there to tag it as the problem with the Instant Client, so I chose "other problem".
                      • 8. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                        user458167
                        Wow, that has been quick. The Oracle support wrote me that this is a known bug (numbered 4612267) with 10.2.0.1 and that I can either receive patch for 10.2.0.1 or wait for the official patchset 10.2.0.2 which should include the fix. So I am going to stay with 10.1.0.4 until 10.2.0.2 becomes available.
                        • 9. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                          BlueR-Oracle
                          Hi,
                          i have the same problem, but in 10.2.0.2. on Linux 2.6.9-42.ELsmp
                          The trace show:
                          .
                          .
                          .

                          open("/opt/oracle/live36/network/admin/sqlnet.ora", O_RDONLY|O_LARGEFILE) = 6
                          fcntl64(6, F_SETFD, FD_CLOEXEC) = 0
                          fstat64(6, {st_mode=S_IFREG|0644, st_size=193, ...}) = 0
                          mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7f13000
                          read(6, "# sqlnet.ora Network Configurati"..., 4096) = 193
                          read(6, "", 4096) = 0
                          close(6) = 0
                          munmap(0xb7f13000, 4096) = 0
                          open("/opt/oracle/live36/rdbms/mesg/ocius.msb", O_RDONLY) = 6
                          fcntl64(6, F_SETFD, FD_CLOEXEC) = 0
                          lseek(6, 0, SEEK_SET) = 0
                          read(6, "\25\23\"\1\23\3\t\t\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 256) = 256
                          lseek(6, 512, SEEK_SET) = 512
                          read(6, "\337y\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 512) = 512
                          lseek(6, 1024, SEEK_SET) = 1024
                          read(6, "\25\7\'\0072\7>\7j\7\276\17$\'\6K5S\24TfT\307T(VsV\222"..., 86) = 86
                          times(NULL) = -2036907799
                          times(NULL) = -2036907799

                          Can you help me because i thank there was a backport of the fixed from 11 to 10.2.0.2 ???
                          • 10. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                            Laurenz Albe
                            I performed a Metalink search and found Bug 5867987 which looks exactly like your problem, reported for 10.2.0.2 on Linux x86.
                            Unfortunately Development seems not to have been able to reproduce the behaviour, as it is closed with status "Could Not Reproduce. To Filer".

                            Is your uptime greater than 248 days?

                            What I would do in your place:
                            - Try again with the latest version of Instant Client (10.2.0.4).
                            - If the problem persists and you have a support contract, open a service request and point out the similarity to bug 5867987.

                            Yours,
                            Laurenz Albe
                            • 11. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                              BlueR-Oracle
                              Hi,
                              thank's so much for your informations.

                              no, the uptime is < at 248 days

                              I suppose the better is to test with 10.2.0.4.

                              Best regards.
                              • 12. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                                750960
                                I can confirm that this is still occurring with 10.2.0.4 on Red Hat AS 4 R 5

                                It appears that on a Linux kernel 2.6 and up the uptime is reduced to 199 days before a failure.

                                My uptime was 211 days, a reboot fixed the problem.
                                • 13. Re: SQL*Plus 10.2.0.1.0 does not work on RHEL3
                                  791601
                                  I've just updated the Oracle instant client to version 10.2.0.4.0 . Can't it fix the problem?