On Fri, Aug 14, 2009 at 03:46:54PM -0400, Stephen Gallagher wrote:
This timeout specifies the lifetime of a cache entry before it is
updated out-of-band. When this timeout is hit, the request will
still complete from cache, but the SSSD will also go and update
the cached entry in the background to extend the life of the
cache entry and reduce the wait time of a future request.
Support for the EnumCacheNoWaitRefreshTimeout is still forthcoming, but
I wanted to get a formal review on this portion.
NACK. I think this patch indicates that nsssrv_cmd.c needs some
refactoring, please do this before adding more code.
More comments follow below.
bye,
Sumit
From c8d774ee2741c76c4c2a07bcae112924b0061e86 Mon Sep 17 00:00:00
2001
From: Stephen Gallagher <sgallagh(a)redhat.com>
Date: Fri, 14 Aug 2009 08:59:53 -0400
Subject: [PATCH] Add support for the EntryCacheNoWaitRefreshTimeout
This timeout specifies the lifetime of a cache entry before it is
updated out-of-band. When this timeout is hit, the request will
still complete from cache, but the SSSD will also go and update
the cached entry in the background to extend the life of the
cache entry and reduce the wait time of a future request.
---
server/examples/sssd.conf | 13 ++
server/man/sssd.conf.5.xml | 13 ++
server/responder/nss/nsssrv.c | 20 ++++
server/responder/nss/nsssrv.h | 7 +-
server/responder/nss/nsssrv_cmd.c | 227 ++++++++++++++++++++++++++++++-------
5 files changed, 236 insertions(+), 44 deletions(-)
diff --git a/server/responder/nss/nsssrv.c b/server/responder/nss/nsssrv.c
index 456c629..6d7bf74 100644
--- a/server/responder/nss/nsssrv.c
+++ b/server/responder/nss/nsssrv.c
@@ -103,6 +103,26 @@ static int nss_get_config(struct nss_ctx *nctx,
&nctx->neg_timeout);
if (ret != EOK) goto done;
+ ret = confdb_get_int(cdb, nctx, NSS_SRV_CONFIG,
+ "EnumCacheNoWaitRefreshTimeout", 0,
+ &nctx->enum_cache_refresh_timeout);
+ if (ret != EOK) goto done;
+ if (nctx->enum_cache_refresh_timeout >= nctx->enum_cache_timeout) {
+ DEBUG(0,("Configuration error: EnumCacheNoWaitRefreshTimeout exceeds"
+ "EnumCacheTimeout. Disabling feature.\n"));
+ nctx->enum_cache_refresh_timeout = 0;
+ }
+
+ ret = confdb_get_int(cdb, nctx, NSS_SRV_CONFIG,
+ "EntryCacheNoWaitRefreshTimeout", 0,
+ &nctx->cache_refresh_timeout);
+ if (ret != EOK) goto done;
+ if (nctx->cache_refresh_timeout >= nctx->cache_timeout) {
+ DEBUG(0,("Configuration error: EntryCacheNoWaitRefreshTimeout
exceeds"
+ "EntryCacheTimeout. Disabling feature.\n"));
+ nctx->cache_refresh_timeout = 0;
+ }
+
ret = confdb_get_string_as_list(cdb, tmpctx, NSS_SRV_CONFIG,
"filterUsers", &filter_list);
if (ret == ENOENT) filter_list = NULL;
I haven't check other timeouts so far, but I think it makes sense to
check if *_timeout < 0.
diff --git a/server/responder/nss/nsssrv.h
b/server/responder/nss/nsssrv.h
index 0d3124c..e756384 100644
--- a/server/responder/nss/nsssrv.h
+++ b/server/responder/nss/nsssrv.h
@@ -50,11 +50,14 @@ struct getent_ctx;
struct nss_ctx {
struct resp_ctx *rctx;
- int cache_timeout;
- int neg_timeout;
struct nss_nc_ctx *ncache;
+ int neg_timeout;
+ int cache_timeout;
int enum_cache_timeout;
+ int cache_refresh_timeout;
+ int enum_cache_refresh_timeout;
+
time_t last_user_enum;
time_t last_group_enum;
The *_timeout variables are use to compare against unsigned values. I
would prefer the *_timeout having a type of time_t or uint*.
diff --git a/server/responder/nss/nsssrv_cmd.c
b/server/responder/nss/nsssrv_cmd.c
index e8f178a..f00a423 100644
--- a/server/responder/nss/nsssrv_cmd.c
+++ b/server/responder/nss/nsssrv_cmd.c
@@ -273,12 +273,15 @@ static void nss_cmd_getpwnam_callback(void *ptr, int status,
struct cli_ctx *cctx = cmdctx->cctx;
struct sss_domain_info *dom;
struct nss_ctx *nctx;
- int timeout;
+ int timeout, refresh_timeout;
see above and
http://freeipa.org/page/Coding_Style#Declaring .
+ time_t now;
uint64_t lastUpdate;
uint8_t *body;
size_t blen;
bool call_provider = false;
bool neghit = false;
+ bool need_callback = true;
+ sss_dp_callback_t cb = NULL;
int ncret;
int ret;
@@ -296,16 +299,33 @@ static void nss_cmd_getpwnam_callback(void *ptr, int status,
if (dctx->check_provider) {
switch (res->count) {
case 0:
+ /* This is a cache miss. We need to get the updated user information
+ * before returning it.
+ */
call_provider = true;
+ need_callback = true;
break;
case 1:
timeout = nctx->cache_timeout;
-
+ refresh_timeout = nctx->cache_refresh_timeout;
lastUpdate = ldb_msg_find_attr_as_uint64(res->msgs[0],
SYSDB_LAST_UPDATE, 0);
- if (lastUpdate + timeout < time(NULL)) {
+ now = time(NULL);
+ if (lastUpdate + timeout < now) {
+ /* This is a cache miss. We need to get the updated user
+ * information before returning it.
+ */
+ call_provider = true;
+ need_callback = true;
+ }
+ else if (refresh_timeout && (lastUpdate + refresh_timeout < now))
{
+ /* We're past the the cache refresh timeout
+ * We'll return the value from the cache, but we'll also
+ * queue the cache entry for update out-of-band.
+ */
call_provider = true;
+ need_callback = false;
}
I would prefer an else-block here. Although call_provider and
need_callback are initialised to do the right thing an else-block would
make it easier to follow the logic here.
break;