请帮忙解释下数据库故障时的这个性能截图

Question

数据库故障时点（16点），网络IO等待很高，处理器队列很长，处理器利用率低，网络流量监视发现此时点的流量接近峰值。这可不可以说明是网络原因导致数据库瓶颈进而发送故障。怎么解释这三条曲线的关系

PS:从服务器到客户端为4M带宽，但不是数据库专用。

Answer 1

Means sessions were wating on network. What kind of error did you get in sql? What kind of processes did machine have? Can't help without details.

Answer 2

网络负荷超重。。估计是某些请求大量数据所致<请求数据量小而并发大的可能性极少，那通常会反映在CPU上>

4点多点这个时间应该是在做什么定时处理，或批任务

---

Try SQL Server 2008 QQ:315054403 [e-mail]

Answer 3

Means sessions were wating on network. What kind of error did you get in sql? What kind of processes did machine have? Can't help without details.

服务器DELL R910，EMC CX240存储，当时连接数800左右。当时的Profiler并未跟踪到大批量查询，可能大批量查询并未结束。请问如何确认问题根源

Answer 4

How much memory does sql has? What's os version? Is it 32 or 64-bit machine? What's sql version and edition? Is clr enabled in sql? Looks like there were memory leak per above error log, any related message in windows event logs? By the way, 800 connections over 4m bandwidth sounds a lot.

Answer 5

How much memory does sql has? What's os version? Is it 32 or 64-bit machine? What's sql version and edition? Is clr enabled in sql? Looks like there were memory leak per above error log, any related message in windows event logs? By the way, 800 connections over 4m bandwidth sounds a lot.

128 GB内存，windows 2008 R2 64 Bit, sql 2008 ent sp1.内存泄露是在重启数据库服务时抛出的。系统日志未报相关错误，带宽是8M,一般情况我们的连接是在600-700左右，有时到900也没出现类似问题。
请问网络异常有没可能造成这种问题，目前我们准备先解决网络带宽专用的问题

Answer 6

High network i/o may cause logon timeout but shouldn't shut down sql service, must be something else was going on. Did you see any unusual alert in windows app event log before stoping sql service? Any other error in sql server log?

Answer 7

High network i/o may cause logon timeout but shouldn't shut down sql service, must be something else was going on. Did you see any unusual alert in windows app event log before stoping sql service? Any other error in sql server log?

没有其他错误，数据库重启是我们人为操作的，因为当时数据库已经很慢，而且新连接无法建立，重启后恢复正常

Answer 8

High network i/o may cause logon timeout but shouldn't shut down sql service, must be something else was going on. Did you see any unusual alert in windows app event log before stoping sql service? Any other error in sql server log?

没有其他错误，数据库重启是我们人为操作的，因为当时数据库已经很慢，而且新连接无法建立，重启后恢复正常

 

32位这种情况遇到过，64位的未发现过，而且你硬件的配置也算不错了。

能不能贴一下较为完整的error log ?

Answer 9

你先看看sql的waits

---

想不想时已是想，不如不想都不想。

Answer 10

High network i/o may cause logon timeout but shouldn't shut down sql service, must be something else was going on. Did you see any unusual alert in windows app event log before stoping sql service? Any other error in sql server log?

没有其他错误，数据库重启是我们人为操作的，因为当时数据库已经很慢，而且新连接无法建立，重启后恢复正常

 

32位这种情况遇到过，64位的未发现过，而且你硬件的配置也算不错了。

能不能贴一下较为完整的error log ?

SQL Server 日志除之前贴出的，没有任何其他的信息或错误。请问你们32位的是什么情况，后来怎么解决的，谢谢

Answer 11

最大线程数 有设置吗？ 多少cpu？ dell r910 是32个cpu吧 ，你这个配置的最大线程数 是 1000左右 ？

有错误号 是 17883 ，17884 吗？  如果是 把 线程数调大 。

Answer 12

最大线程数 有设置吗？ 多少cpu？ dell r910 是32个cpu吧 ，你这个配置的最大线程数 是 1000左右 ？

有错误号 是 17883 ，17884 吗？  如果是 把 线程数调大 。

没有这方面错误

Answer 13

还是建议 调大 线程池  服务器现象跟线程池满很类似；还有 那两个是事件号（17883,17884） 不是错误我说错了。

Answer 14

High network i/o may cause logon timeout but shouldn't shut down sql service, must be something else was going on. Did you see any unusual alert in windows app event log before stoping sql service? Any other error in sql server log?

没有其他错误，数据库重启是我们人为操作的，因为当时数据库已经很慢，而且新连接无法建立，重启后恢复正常

 

32位这种情况遇到过，64位的未发现过，而且你硬件的配置也算不错了。

能不能贴一下较为完整的error log ?

SQL Server 日志除之前贴出的，没有任何其他的信息或错误。请问你们32位的是什么情况，后来怎么解决的，谢谢

32位多是memtoleave内存不足造成的，可以使用-g增大该值，但64位怎么会呢？

当时如果能通过sys.dm_os_memory_clerks抓下内存使用情况就好了.