I keep losing my connection to hashcows constantly. Is this in my end or is this something that happens?
Phad
* edit * - It happens everytime hashcows switches to a new coin. There must be a cgminer setting that prevents this.
Same here, thats why I will use hashcows as a backup server
Well, to avoid 5 minutes stability check you could go back to version 3.5.0 according to procrypt in #hashcows but if you're compiling cgminer yourself, here's a patch that lowers the retry-time after the pool switches coin down to 5s (instead of 30) and then checks stability for 5-8s:
--- cgminer.c 2014-01-29 21:56:32.959805084 +0100
+++ cgminer.c-patched 2014-01-30 09:18:41.807082381 +0100
@@ -5553,7 +5553,7 @@
while (!restart_stratum(pool)) {
if (pool->removed)
goto out;
- cgsleep_ms(30000);
+ cgsleep_ms(5000);
}
stratum_resumed(pool);
continue;
@@ -7083,19 +7083,19 @@
pool->testing = false;
}
- /* Test pool is idle once every minute */
- if (pool->idle && now.tv_sec - pool->tv_idle.tv_sec > 30) {
+ /* Test if pool is idle. */
+ if (pool->idle) {
cgtime(&pool->tv_idle);
if (pool_active(pool, true) && pool_tclear(pool, &pool->idle))
pool_resus(pool);
}
/* Only switch pools if the failback pool has been
- * alive for more than 5 minutes to prevent
+ * alive for more than 5 seconds to prevent
* intermittently failing pools from being used. */
if (!pool->idle && pool_strategy == POOL_FAILOVER && pool->prio < cp_prio() &&
- now.tv_sec - pool->tv_idle.tv_sec > 300) {
- applog(LOG_WARNING, "Pool %d %s stable for 5 mins",
+ now.tv_sec - pool->tv_idle.tv_sec > 5) {
+ applog(LOG_WARNING, "Pool %d %s stable for >5 seconds.",
pool->pool_no, pool->rpc_url);
switch_pools(NULL);
}
@@ -7109,7 +7109,7 @@
switch_pools(NULL);
}
- cgsleep_ms(30000);
+ cgsleep_ms(2500);
}
return NULL;
On a sidenote, if you're running GNU/Linux and allready have all dependencies installed and ADL_SDK downloaded and unzipped to ~/AMD/ADL_SDK you'd download, patch, configure and compile by something like this:
And 'make install' if you want to.
Edit: This should be a very obvious snippet, but otherwise, if you don't know what's going on it's probably good practice to not patch and run stuff you get from this or any other forum about cryptos until someone trustworthy OK'd it. Pastebin apparantly does something to the end of the file that causes 'patch' to throw an error but the file still gets patched and compiles just fine. Also, i updated the patch to lower the time between retrying to set up a new connection.
I could just skip telling you this and take a large(r) chunk of the fast blocks, but I want to support the exchange that show the most transparancy now that... well, some don't.