lei.chen/myxce
1
0
Fork 1
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

feat(xce-test-e2e-primitives): quiet wait-primary poll — suppress per-attempt joblist perror + libxce failover INFO spam

Browse source 
The wait-primary probe loop EXPECTS repeated failure until a primary serves, so it
no longer spams the console: joblist_query gains a quiet flag (the poll passes
quiet=1) and the client log threshold is raised to warn for the loop duration,
restored to the user's level on every exit path. Interactive joblist + the
assertions keep their diagnostics (quiet=0).
This commit is contained in:
Rongsong Shen 2026-06-20 03:18:40 +00:00
parent c205b5b25d
commit 2988682bd6
1 changed files with 30 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

36
src/controller/testsuite/test.c
View file

@ -901,7 +901,8 @@ static char* job_state_str(int state)
*
* On error this returns the negative rc WITHOUT setting any auto-variable, so a
* failed joblist never advertises a fake `$jobcount` (Req 3.6). */
static int joblist_query(const char* find_jobid, int* contains, int print_table)
static int joblist_query(const char* find_jobid, int* contains, int print_table,
int quiet)
{
int rc, i;
xce_job_t** jobs;
@ -914,7 +915,12 @@ static int joblist_query(const char* find_jobid, int* contains, int print_table)
if (rc < 0) {
SET_ERROR(-rc);
perror("xce_joblist");
/* `quiet` callers (e.g. the wait-primary poll, which probes repeatedly
* and EXPECTS failure until a primary serves) suppress the per-attempt
* error print so a normal failover wait does not spam the console. */
if (!quiet) {
perror("xce_joblist");
}
return rc;
}
@ -957,7 +963,7 @@ int cmd_joblist(int argc, char* argv[])
{
int count;
count = joblist_query(NULL, NULL, 1);
count = joblist_query(NULL, NULL, 1, /*quiet=*/0);
if (count < 0) {
return count;
@ -1008,7 +1014,7 @@ int cmd_expect_jobs(int argc, char* argv[])
/* Print-free count via the shared helper. A negative rc means the query
* itself failed (e.g. no reachable master) — that is an assertion failure,
* not "0 jobs". */
count = joblist_query(NULL, NULL, 0);
count = joblist_query(NULL, NULL, 0, /*quiet=*/0);
if (count < 0) {
printf("expect-jobs: could not determine job count (joblist query "
@ -1056,7 +1062,7 @@ int cmd_expect_job(int argc, char* argv[])
/* Print-free membership query via the shared helper. A negative rc means the
* query itself failed (e.g. no reachable master) — that is an assertion
* failure, distinct from "job absent". */
rc = joblist_query(argv[1], &contains, 0);
rc = joblist_query(argv[1], &contains, 0, /*quiet=*/0);
if (rc < 0) {
printf("expect-job: could not determine membership of '%s' (joblist "
@ -1099,6 +1105,11 @@ int cmd_expect_job(int argc, char* argv[])
#define NS_PER_MS 1000000L
#define US_PER_MS 1000L
/* The client's configured log level (set in main from -v), so wait-primary can
* temporarily raise the threshold to silence libxce's per-attempt failover INFO
* spam during its poll and then RESTORE the user's chosen level. */
static char* g_active_loglevel = (char*)"info";
static long monotonic_ms(void)
{
struct timespec ts;
@ -1135,15 +1146,23 @@ int cmd_wait_primary(int argc, char* argv[])
start_ms = monotonic_ms();
/* The poll probes repeatedly and EXPECTS failure until a primary serves, so
* it must not spam the console: the probe suppresses its own error print
* (quiet=1), and we raise the client log threshold for the duration of the
* loop to silence libxce's per-attempt "xcecli failover: attempt N" INFO
* logs — RESTORED to the user's level on every exit path. */
mylog_setlevel((char*)"warn");
/* Probe, then re-check the monotonic deadline between attempts. Each probe
* may itself take time (libxce failover has its own internal retry budget);
* we re-read the clock after every probe returns so the overall loop honors
* the deadline and returns failure in bounded time. */
for (;;) {
int rc = joblist_query(NULL, NULL, 0);
int rc = joblist_query(NULL, NULL, 0, /*quiet=*/1);
if (rc >= 0) {
elapsed_ms = monotonic_ms() - start_ms;
mylog_setlevel(g_active_loglevel);
printf("wait-primary: primary serving after %ld ms\n", elapsed_ms);
return 0;
}
@ -1151,6 +1170,7 @@ int cmd_wait_primary(int argc, char* argv[])
elapsed_ms = monotonic_ms() - start_ms;
if (elapsed_ms >= timeout_ms) {
mylog_setlevel(g_active_loglevel);
printf("wait-primary: no primary served within %ld ms (waited "
"%ld ms)\n",
timeout_ms, elapsed_ms);
@ -3079,6 +3099,10 @@ int main(int argc, char* argv[])
loglevel = verbose_to_loglevel(arguments.verbose);
}
/* Remember the user's level so wait-primary can restore it after temporarily
* raising the threshold to silence its poll-loop failover spam. */
g_active_loglevel = loglevel;
mylog_init(logfile, loglevel, logsize, logcount);
signal(SIGINT, sigbreak);