[dev.icinga.com #2503] 1.7.0dev test issues on solaris

[icinga-migration]

This issue has been migrated from Redmine: https://dev.icinga.com/issues/2503

Created by Tommi on 2012-04-06 12:46:57 +00:00

Assignee: Tommi
Status: Resolved (closed on 2012-04-26 19:01:14 +00:00)
Target Version: 1.7
Last Update: 2014-12-08 14:37:38 +00:00 (in Redmine)

Icinga Version: 1.10.0
OS Version: any

---

ido2db on oracle is not working. It terminates without core file shortly after start and leave the process as zombie similar #2458. i will send the logs seperate as private mail. Icinga itself looks got from the first hour.

Build raises some warnings in solaris with system supplied cc

./configure --prefix=xxx/icinga \
--with-icinga-user=adbm \
--with-icinga-group=adbm \
--with-web-user=adbm \
--with-web-group=adbm \
--with-command-user=adbm \
--with-command-group=adbm \
--enable-embedded-perl \
--enable-nanosleep \
--enable-idoutils \
--enable-event-broker \
--with-httpd-conf=xxx/Apache2/conf \
--with-temp-dir=xxx/prod/monitor/tmp \
--with-htmlurl=/monitor \
--enable-ssl \
--with-ssl=xxx \
--with-init-dir=xxx/etc/init.d \
--localstatedir=xxx/prod/monitor/var \
--sysconfdir=xxx/prod/monitor/etc \
--enable-oracle \
--enable-nagiosenv \
--with-ocilib-lib=xxx/lib \
--with-ocilib-inc=xxx/include \
--with-oracle-lib=xxx/instantclient64_11_2

configure: WARNING: unrecognized options: --with-htmlurl

Make all Warnings:

"nebmods.c", line 261: warning: assignment type mismatch:
        pointer to function(int, pointer to char, pointer to void) returning
int "=" pointer to void
"nebmods.c", line 327: warning: assignment type mismatch:
        pointer to function(int, int) returning int "=" pointer to void
"nebmods.c", line 560: warning: assignment type mismatch:
        pointer to function(int, pointer to void) returning int "=" pointer
to void

"events.c", line 954: warning: assignment type mismatch:
        pointer to function(void) returning long "=" pointer to void
"events.c", line 1617: warning: assignment type mismatch:
        pointer to function(pointer to void) returning void "=" pointer to
void
"events.c", line 1881: warning: assignment type mismatch:
        pointer to function(void) returning long "=" pointer to void
"events.c", line 1902: warning: assignment type mismatch:
        pointer to function(void) returning long "=" pointer to void
"getcgi.c", line 44: warning: loop not entered at top

statuswml.c:
../base/statsprofiler.c:
ld: warning: symbol `dummy' has differing sizes:
        (file statuswml.o value=0x8; file macros-cgi.o value=0x4);
        statuswml.o definition taken
ld: warning: symbol `dummy' has differing sizes:
        (file statuswml.o value=0x8; file statusdata-cgi.o value=0x4);
        statuswml.o definition taken
ld: warning: symbol `dummy' has differing sizes:
        (file statuswml.o value=0x8; file downtime-cgi.o value=0x4);
        statuswml.o definition taken

statusmap.c:
"statusmap.c", line 2279: warning: loop not entered at top
../base/statsprofiler.c:
"helloworld.c", line 38: warning: syntax error:  empty declaration

"../../../common/snprintf.c", line 285: warning: initializer does not fit or
is out of range: -1

"../include/sla.h", line 34: warning: zero or negative subscript
"../include/sla.h", line 57: warning: zero or negative subscript

"ido2db.c", line 1129: warning: statement not reached
"ido2db.c", line 1177: warning: statement not reached

Changesets

2012-04-09 08:48:29 +00:00 by Tommi 83671d3

idoutils: add several null pointer checks before handover strings to debuglog #2503
refs #2503, #2271

2012-04-28 09:20:08 +00:00 by Tommi 3fe913f

idoutils: add several null pointer checks before handover strings to debuglog #2503
refs #2503, #2271

Conflicts:

	Changelog

---

Relations:

• relates #2271
• relates #2500
• relates #2503

[icinga-migration]

Updated by Tommi on 2012-04-08 10:08:02 +00:00

looks like the vfprintf error from #2271 (and others) is back again

(dbx) run -f -c etc/ido2db-oracle.cfg
Running: ido2db -f -c etc/ido2db-oracle.cfg
(process id 10207)
Reading libociei.so
t@1 (l@1) signal SEGV (no mapping at the fault address) in strlen at
0xffffffff7e139c70
0xffffffff7e139c70: strlen+0x0050:      ld       [%o2], %o1
Current function is ido2db_log_debug_info
  145           vfprintf(ido2db_debug_file_fp, fmt, ap);
(dbx) where
current thread: t@1
  [1] strlen(0x0, 0x53, 0x0, 0x0, 0x0, 0x53), at 0xffffffff7e139c70
  [2] _ndoprnt(0x10004fd5d, 0xffffffff7fffe160, 0xffffffff7e1a11e4,
0x1004b505a, 0x0, 0x10004fd5c), at 0xffffffff7e1a2f54
  [3] vfprintf(0xffffffff7e2f0428, 0x10004fd30, 0xffffffff7fffe158,
0x14106c, 0x140c08, 0xffffffff7e1a508c), at 0xffffffff7e1a54c0
=>[4] ido2db_log_debug_info(level = 1, verbosity = 2, fmt = 0x10004fd30
"ido2db_query_objects_select_name1_name2(%s,%s) object id=%lu selected\n",
... = 0xffffffff00000000, ...), line 145 in "logging.c"
  [5] ido2db_get_object_id(idi = 0xffffffff7fffeaf0, object_type = 12, n1 =
0x1005452e0 "check_nrpe", n2 = (nil), object_id = 0xffffffff7fffe4c8), line
248 in "dbhandlers.c"
  [6] ido2db_get_object_id_with_insert(idi = 0xffffffff7fffeaf0, object_type
= 12, n1 = 0x1005452e0 "check_nrpe", n2 = (nil), object_id =
0xffffffff7fffe4c8), line 308 in "dbhandlers.c"
  [7] ido2db_handle_servicecheckdata(idi = 0xffffffff7fffeaf0), line 2667 in
"dbhandlers.c"
  [8] ido2db_end_input_data(idi = 0xffffffff7fffeaf0), line 2145 in
"ido2db.c"
  [9] ido2db_handle_client_input(idi = 0xffffffff7fffeaf0, buf = 0x10049b330
"999"), line 1813 in "ido2db.c"
  [10] ido2db_check_for_client_input(idi = 0xffffffff7fffeaf0), line 1485 in
"ido2db.c"
  [11] ido2db_handle_client_connection(sd = 7), line 1363 in "ido2db.c"
  [12] ido2db_wait_for_connections(), line 1165 in "ido2db.c"
  [13] main(argc = 4, argv = 0xffffffff7ffff7c8), line 312 in "ido2db.c"
(dbx) display fmt
fmt = 0x10004fd30 "ido2db_query_objects_select_name1_name2(%s,%s) object
id=%lu selected\n"
(dbx) whatis ap
va_list ap;

[icinga-migration]

Updated by Tommi on 2012-04-09 09:07:11 +00:00

• Category set to 71
• Status changed from New to Feedback
• Assigned to set to Tommi
• Target Version set to 1.7
• Done % changed from 0 to 100

added several null pointer checks in changeset 83671d3, which solves the startup problem up to now

[icinga-migration]

Updated by mfriedrich on 2012-04-19 13:06:06 +00:00

would be good if carl could have a look. i don't have ido on solaris.

[icinga-migration]

Updated by crfriend on 2012-04-22 23:47:18 +00:00

dnsmichi wrote:

would be good if carl could have a look. i don't have ido on solaris.

It's nice to be remembered. It's also worth noting that I do not have access to an Oracle system, so here I'm flying in the dark.

From the looks of this one -- especially during startup where there will necessarily be some data left incomplete or unknown (read, "NULL"), we should expect these sorts of calls. Specifically, in this case, it looks like we're populating the commands table in which "n2"/"name2" is VALID as a null; here, the logging routine should flag the case as "unset", output the log entry, and press on.

I advocate for checking for null-pointers in the logging subroutines and flagging those as a string of "(unset)" therein. I DO NOT advocate for doing such things in situations where that string could be passed downstream to other processes or written as a string to the database. Personally, I prefer the notion of using the verbiage of "unset" over "null" as it's more precise in this usage and cannot be confused with anything else.

[icinga-migration]

Updated by Tommi on 2012-04-26 19:01:14 +00:00

• Status changed from Feedback to Resolved

The current solution works fine for me. And yes, i hope to found all possible occurencies of null pointers in logging and replaced it with "(null)"

[icinga-migration]

Updated by mfriedrich on 2014-12-08 14:37:38 +00:00

• Project changed from 18 to Core, Classic UI, IDOUtils
• Category changed from 71 to IDOUtils
• Icinga Version set to 1
• OS Version set to any