Re: Jenkins build became unstable: epics-base-3.15 #98

[Andrew Johnson <anj@aps.anl.gov>]

On 09/22/2014 05:10 PM, APS Jenkins wrote:
> See <https://jenkins.aps.anl.gov/job/epics-base-3.15/98/changes>

Only 6253 tests ran, the previous build had 8089.

Since Jenkins' TAP parser isn't showing the actually failures here's
what I get running the tests manually:

epicsSockResolveTest.t ..... Bailout called.  Further testing stopped:
hostToIPAddr() is broken, testing not possible
FAILED--Further testing stopped: hostToIPAddr() is broken, testing not
possible
tux% ./epicsSockResolveTest
1..33
Bail out! hostToIPAddr() is broken, testing not possible
Abort



arrayOpTest.t ........ Failed 1/10 subtests

tux% ./arrayOpTest
1..10
# Test dbGet() and dbPut() of an array
############################################################################
## EPICS R3.15.0.1-DEV $$Date$$
## EPICS Base built Jul 23 2014
############################################################################
# Fetch initial value
ok  1 - prec->nord==0
ok  2 - nreq==0
ok  3 - memcmp(&addr, &save, sizeof(save))==0
# Write a new value
ok  4 - dbPut(&addr, DBF_DOUBLE, &data, NELEMENTS(data))==0
ok  5 - prec->nord==4
ok  6 - pbtr[0]==4 && pbtr[1]==5 && pbtr[2]==6 && pbtr[3]==7
not ok  7 - memcmp(&addr, &save, sizeof(save))==0
# reread the value
ok  8 - nreq==NELEMENTS(data)
ok  9 - data[0]==4.0 && data[1]==5.0 && data[2]==6.0 && data[3]==7.0
ok 10 - memcmp(&addr, &save, sizeof(save))==0
dbmfFinish: not all free

    Results
    =======
       Tests: 10
      Passed:   9 = 90.00%
      Failed:   1 = 10.00%

However on Jenkins there were more tests failing. Both analogMonitorTest
and arrayOpTest bailed out instead, might have missed one or two others
as well:

Bail out! Error reading test database 'analogMonitorTest.db'


Bail out! Failed to load test database
dbReadDatabase(arrayOpTest.db,.:..:../O.Common:O.Common,(null))


Got to go now, if anyone else can contribute to working out what broke
before I get in tomorrow I would be grateful. It's possible that my
commit #12529 had some effect, although I ran the tests and Jenkins
passed run #97 with that included. Jenkins should be starting from
scratch so builds *should* be independent, but I'm not actually doing a
'make distclean' so that might not be what's actually happening.

- Andrew