Sets the maximum number of workers. That's basically the number of concurrent connections the server is capable to process. Default is 250. There are some limits set by the OS, but hssTVS will try to increase the number of max open filedescriptors accordingly. So there's no need to use ulimit for that anymore.

Maximum size of the receive buffer for http requests. Default is 2500 (bytes), smaller values are ignored. One customer had problems as the typical header contains one or several history cookies which might get pretty large. So it is now possible to increase the buffer size as needed. Bigger buffers will need more memory per connection and it will slow the parsing down. Every connection needs 2 receive buffers (for pipelining purposes) which are pre-allocated at start-up, so don't enlarge it without real need.

Seconds to wait until an idle persistent connection is closed after the last request is finished. Defaults to a setting of 1. The problem is that many browsers use more than 1 connection. For example Opera uses 4 to 8, while Mozilla related ones use normally 2, but often can be set higher. Those remain open, often unused until they time out. For the sake of persistent connections, which can enhance the max number of requests per second the server can handle by up to 200%, you should keep that at least at 1 second. A 0 closes the connection right after a request is finished unless the next is already in the receive buffer (pipelining).

Max number of requests per persistent connection until close is forced, 0 for no max. Defaults to 0. A limit (usually 5 to 20) can help reducing the response time on high load servers as well as reduce the effect of a possible abuse of persistent connections.

This is a flag and takes no value. The newer LRU cache subsystem is more flexible. It creates stat_entries in the cache memory as needed. If this flag is not set, the sLRU will be used (old one) and you might need to specify the number of just stat entries.

This value determins how big a file might be(in bytes) to be added to the cache. While we see a major speed up for small files that reside in the cache, the advantage diminishes when the files get bigger. So, to make the most out of the ram you set aside as a cache you have to draw a line to shut out the bigger files. The default is set to 32KByte and anything bigger would be a waste under normal circumstances, 0 disables the file mmaping of the cache.

The maximum size of the cache in MBytes. We have to make sure that we don't exceed the systems real ram at any time. The server has no means to determin when that'll happen until even the swap area is used up completely. The moment the system requires to swap memory in and out, the moment the whole system suffers a major slow down. Another fact is that you are limited to around 2GByte address space per process on a 32Bit engine, no matter how much (virtual)ram you really have. The default setting is 50 MBytes, 0 disables the file mmaping of the cache.

This value determins how long an idle file entry will stay in the cache (in seconds). It also kinda limits the numbers of files that fill the cache manger up, so it might speed things a bit up if we get rid of the idle ones after a time. The default is off, the maximum is 1 day, 0 disables the cleaner.

This entry specifies the time intervall in seconds cached files are checked against the original on the disk to catch updates. The default is 60 seconds. A high number will give better performance but will also make hssTVS fail to recognice changed files.

This setting is only needed/ used for sLRU, LRU stat entries reside in the cache ram. The cache works in 2 ways. The main mode caches a file's status data and mmaps the file into memory. But in case the file is too big, there is not enough space left or if the mmapping is disabled, only the status data is stored and parts of its header pre-created. This value limits the number of such entries. The default (and minimum) is to use twice the number of max_connections. hssTVS will increase the number if set too low.

This is a flag and takes no value. This will disable the mmapping functionallity of the cache subsystem. File status information is still stored though, giving a bit more speed (not overwhelming though). Only the just_stat_entry_max is used, the other cache related values are ignored. This will drastically slow the system down, so be sure you really want it that way and don't complain afterwards.

This is a flag and takes no value. If you put this into the configuration the cache will be completely disabled. No files will be mapped and no status information stored. The whole thing is just bypassed. Of course all other cache related settings will be ignored then and it does overrule the "disable mmap" setting. However, this will drastically slow the system down, so be sure you really want it that way and don't complain afterwards.

Defines the time in seconds the server checks for idle cache entries and workers. A short intervall will use up more CPU time but will give a more accurate handling of the involved timers. Default is 3 seconds.

The values are always the gathered ones since the last time the stats were retrieved. In case you don't want those stats logged in intervalls you can also retrieve them by sending a SIGUSR1 signal ( []kill -SIGUSR1 `cat /var/run/hsstvs.pid` ). This feature is intended for monitoring and tuning.
Warning, to use the signal you have to set stat_intervall to -1, else the numbers will not be correct.