求助：运行程序时出现内存不够，ulimit后仍然显示内存不够，该如何办

titansword2000

chenyx 发表于 2013-03-02 19:39
问题是你的内存使用量不大,应该是程序或者系统的限制.你们以前遇到过类似的问题没有

以前没遇到过的

vermouth

used 这么点，哪提示不够了？先看看耗费到哪了吧

titansword2000

vermouth 发表于 2013-03-02 20:24
used 这么点，哪提示不够了？先看看耗费到哪了吧

就是因为系统显示内存足够，但程序却不能成功分配足够的内存才感到疑惑啊。具体错误提示来自程序运行的日志文件，程序本身非正常退出。

titansword2000

继续等待中...

titansword2000

当使用ltrace进行跟踪时，在程序运行结束前出现如下的记录：


_mp_v(0x230d5924, 0xa1d5b43, 0, 12, 0x7192eeb0)  = 337
memcpy(0x5cddd3e0, "]", 1)                       = 0x5cddd3e0
memcpy(0x21850258, "]", 1)                       = 0x21850258
_mp_p(0x230d5924, 0xa12f92a, 0xbffe4678, 0x43f768, 56) = 352
malloc(352)                                      = 0x68da0a20
_mp_v(0x230d5924, 0xa12f92a, 0xbffe4678, 0x43f768, 56) = 0x68da0a20
_mp_p(0x230d5924, 0xa12f92a, 0xbffe4678, 0x43f768, 56) = 352
_mp_v(0x230d5924, 0xa12f92a, 0xbffe4678, 0x43f768, 56) = 0x68da0a20
_mp_p(0xaed1a60, 32768, 512, 512000, 0xbffe46a0) = 0xbffe49dc
__mth_i_kcmp(0x1819688, 0, 4, 0, 0xbffe46a0)     = 1
_mp_v(0xaed1a60, 0, 4, 0, 0xbffe46a0)            = 0x63d06fc8
_mp_p(0x230d5924, 0, 0, 0, 0)                    = 0x63d06fc8
free(0x63d06fc0)                                 = <void>
_mp_v(0x230d5924, 0, 0, 0, 0)                    = 0x44a84c
_mp_p(0x230d5924, 0x43f78c, 352, 0x218b1f60, 0xbffe466 = 0x18196a0
malloc(25292624)                                 = NULL
_mp_v(0x230d5924, 0x43f78c, 352, 0x218b1f60, 0xbffe466 = 0
_mp_bcs_stdio(0x230d5924, 0x43f78c, 352, 0x218b1f60, 0xbffe466 = 0
sprintf("ALLOCATE: 25269896 bytes request"..., "ALLOCATE: %lu bytes requested; n"..., 25269896) = 53
_mp_ecs_stdio(0xbffe45d4, 0xaed196e, 0x1819688, 0x218b1f60, 0xbffe466 = 53
sprintf("0: ALLOCATE: 25269896 bytes requ"..., "%d: %s\n", 0, "ALLOCATE: 25269896 bytes request"...) = 57
strlen("0: ALLOCATE: 25269896 bytes requ"...)    = 57
write(2, "0: ALLOCATE: 25269896 bytes requ"..., 570: ALLOCATE: 25269896 bytes requested; not enough memory
) = 57
__pgi_abort(1, 0, 0xbffe4578, 0xa1b1f39, 2 <unfinished ...>
+++ exited (status 127) +++
[liuc@127 4dvar]$
[liuc@127 4dvar]$

xuhainanjing

这个程序本身有log么？我不是说ltrace的log！

titansword2000

xuhainanjing 发表于 2013-03-05 09:18
这个程序本身有log么？我不是说ltrace的log！

程序本身也有log。我贴程序本身log的最后部分吧，如下：


====================================
.....................................
In dolph: TIME, W =  28.000000      0.2051719E-05
In dolph: TIME, W =  29.000000      0.5513138E-06
In dolph: TIME, W =  30.000000      0.9622389E-07
In dolph: SUM OF WEIGHTS W2 =     1.0000
INPUT LandUse = "USGS"
  LANDUSE TYPE = "USGS" FOUND           33  CATEGORIES            2  SEASONS WATER CATEGORY =            16  SNOW CATEGORY =            24
Timing for processing lateral boundary for domain        1:    0.03873 elapsed seconds
d01 2008-02-05_12:00:00 Push xtraj time_stamp:2008-02-05_12:00:00
WRF TILE   1 IS      1 IE     90 JS      1 JE     60
WRF NUMBER OF TILES =   1
Timing for main: time 2008-02-05_12:06:00 on domain   1:   16.59303 elapsed seconds
d01 2008-02-05_12:06:00 Push xtraj time_stamp:2008-02-05_12:06:00
Timing for main: time 2008-02-05_12:12:00 on domain   1:   12.43550 elapsed seconds
d01 2008-02-05_12:12:00 Push xtraj time_stamp:2008-02-05_12:12:00
Timing for main: time 2008-02-05_12:18:00 on domain   1:   12.57382 elapsed seconds
d01 2008-02-05_12:18:00 Push xtraj time_stamp:2008-02-05_12:18:00
Timing for main: time 2008-02-05_12:24:00 on domain   1:   12.62680 elapsed seconds
d01 2008-02-05_12:24:00 Push xtraj time_stamp:2008-02-05_12:24:00
Timing for main: time 2008-02-05_12:30:00 on domain   1:   12.53811 elapsed seconds
d01 2008-02-05_12:30:00 Push xtraj time_stamp:2008-02-05_12:30:00
Timing for main: time 2008-02-05_12:36:00 on domain   1:   16.96066 elapsed seconds
d01 2008-02-05_12:36:00 Push xtraj time_stamp:2008-02-05_12:36:00
Timing for main: time 2008-02-05_12:42:00 on domain   1:   12.55948 elapsed seconds
d01 2008-02-05_12:42:00 Push xtraj time_stamp:2008-02-05_12:42:00: ALLOCATE: 25269896 bytes requested; not enough memory
[liuc@127 4dvar]$


由其log可见，分配内存失败导致程序退出。

titansword2000

有知道解决这问题的大拿吗

linuxforlive

我同意三楼
网上搜一下如何将一个文件作为swap使用
下面的命令可能有记不清的了 最好自己网上确认一下
ddif =/dev/zero of=xxx bs=1M count=自己定
mkswap xxx
swapon xxx