From: Dmitry K. <di...@de...> - 2001-12-04 11:55:25
|
Hello, All! I found lost gfix switch. If I'm crazy or completely wrong - tell me please. (I tried all the following with 5.6, and got wrong results. I don't understand why I'm sure that all worked in 5.6). gfix -cache gfix's help shows that this switch used to "shutdown cache manager". And, I remember it worked (I've used it on training to show how much page reads is made by executing query. Main idea was that -cache clears superserver cache, and new query will read all needed pages from disk, showing it in isc_database_info statistics). Now it doesn't work. Trying to gfix database.gdb -cache gives "Invalid switch combination". Documentation lookup (beta docs 6.0 and release 6.0 docs) showed that -cache n Reserved for future implementation. huh! (looking in to 5.x doc I found the same :-( ) Okay, I'm cool and can look into sources. aliceswi.h is Ok, and have line IN_SW_ALICE_CACHE, 0, "cache", sw_cache, sw_shut, 0, FALSE, 30, 0, NULL, As I understand, sw_shut here indicates that sw_cache is a sub-option for sw_shut. This is NONSENSE, I think. I can't imagine how -cache can tie with -shut. And, really, calling gfix database.gdb -shut -cache 0 turning database to "shutdown" state. Nothing more. Of course, because alice.c have if (table->in_sw_value & (sw_attach | sw_force | sw_tran | sw_cache)) { setting number of seconds interval for ALICE_data.ua_shutdown_delay} and if((switches & sw_shut) && !(switches & ((sw_attach | sw_force | sw_tran | sw_cache)))) ALICE_error (19,0,0,0,0,0); /* msg 19: must specify type of shutdown */ Most funny thing is that structure ALICE_data does not have anything for cache, and -cache itself does not being processed in any way. Seems that somebody wiped out part of code and changed another part in a strange way? Or -cache never worked as a separate option, and I have wrong memories? -- Dmitry Kuzmenko, Epsylon Technologies. |