On 10/5/18 8:34 AM, Michael Trapp wrote:
Vhostmd uses the update_interval value to sleep() between the metrics updates. But a long runtime of the metrics updates can result in a noticeable longer time between the updates and a consumer might read 'old' data after waiting update_interval seconds. Fix the delay between updates, that 'runtime of metrics update' + 'sleep time' matches the update_interval value. --- vhostmd/vhostmd.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)
Reviewed-by: Jim Fehlig <[email protected]> Regards, Jim
diff --git a/vhostmd/vhostmd.c b/vhostmd/vhostmd.c index 1718df2..9ce43f4 100644 --- a/vhostmd/vhostmd.c +++ b/vhostmd/vhostmd.c @@ -969,6 +969,9 @@ static int vhostmd_run(int diskfd) }while (!down) {+ time_t run_time, + start_time = time(NULL); + vu_buffer_add(buf, "<metrics>\n", -1); if (metrics_host_get(buf)) vu_log(VHOSTMD_ERR, "Failed to collect host metrics " @@ -987,7 +990,11 @@ static int vhostmd_run(int diskfd) #endif if (ids) free(ids); - sleep(update_period); + + run_time = time(NULL) - start_time; + if ((run_time > 0) && (run_time < update_period)) + sleep(update_period - run_time); + vu_buffer_erase(buf); } vu_buffer_delete(buf);
_______________________________________________ virt-tools-list mailing list [email protected] https://www.redhat.com/mailman/listinfo/virt-tools-list
