Hi Vu,
My comments:
- The max_waiting_time_10s, do we really need a time limitation? The clm
initialization should be completed, now it is threaded, so it can retry
till success
- As same as Lennart's comment on #2192, use base::Sleep, or osaf_nanosleep
Thanks,
Minh
On 17/11/16 17:28, Vu Minh Nguyen wrote:
> osaf/services/saf/ntfsv/ntfs/ntfs_clm.c | 14 +++++++++++++-
> 1 files changed, 13 insertions(+), 1 deletions(-)
>
>
> NTF did not deal with TRY_AGAIN error code of `saClmInitialize()`,
> NTF would exit, and cause node reboot if getting TRY_AGAIN.
>
> The patch adds a while loop to do retry when getting TRY_AGAIN.
>
> diff --git a/osaf/services/saf/ntfsv/ntfs/ntfs_clm.c
> b/osaf/services/saf/ntfsv/ntfs/ntfs_clm.c
> --- a/osaf/services/saf/ntfsv/ntfs/ntfs_clm.c
> +++ b/osaf/services/saf/ntfsv/ntfs/ntfs_clm.c
> @@ -101,13 +101,25 @@ void *ntfs_clm_init_thread(void *cb)
> {
> ntfs_cb_t *_ntfs_cb = (ntfs_cb_t *) cb;
> SaAisErrorT rc = SA_AIS_OK;
> + uint32_t msecs_waited = 0;
> + const uint32_t max_waiting_time_10s = 10 * 1000; /* 10 secs */
> +
> TRACE_ENTER();
> +
> rc = saClmInitialize_4(&_ntfs_cb->clm_hdl, &clm_callbacks, &clmVersion);
> + while (((rc == SA_AIS_ERR_TRY_AGAIN) || (rc == SA_AIS_ERR_TIMEOUT) ||
> + (rc == SA_AIS_ERR_UNAVAILABLE)) &&
> + (msecs_waited < max_waiting_time_10s)) {
> + usleep(100*1000);
> + msecs_waited += 100;
> + rc = saClmInitialize_4(&_ntfs_cb->clm_hdl, &clm_callbacks,
> &clmVersion);
> + }
> if (rc != SA_AIS_OK) {
> LOG_ER("saClmInitialize failed with error: %d", rc);
> TRACE_LEAVE();
> - exit(EXIT_FAILURE);
> + exit(EXIT_FAILURE);
> }
> +
> rc = saClmSelectionObjectGet(_ntfs_cb->clm_hdl,
> &ntfs_cb->clmSelectionObject);
> if (rc != SA_AIS_OK) {
> LOG_ER("saClmSelectionObjectGet failed with error: %d", rc);
>
------------------------------------------------------------------------------
_______________________________________________
Opensaf-devel mailing list
Opensaf-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-devel
