我无法找到 503 错误的原因。我怀疑内存缓存有问题，但在 strace 中似乎出现“zend_mm_heap 损坏”错误。
这是配置文件
<FilesMatch ".+\.php$">
SetHandler "proxy:unix:/run/php/php-flm.sock|fcgi://flm-https"
</FilesMatch>
<Proxy "fcgi://flm-https" timeout=300 min=16>
</Proxy>
和一些来自 strace 的日志
[pid 487992] 13:13:40 fstat(7, {st_mode=S_IFREG|0644, st_size=172, ...}) = 0 <0.000023>
[pid 487992] 13:13:40 read(7, "<?php\n\nnamespace CSC\\Cache;\n\nuse CSC\\Common\\CacheConstants;\n\nclass UserSessionCache extends AbstractCache\n{\n protected static $cacheKey = CacheConstants::USER_SESSION;\n}", 172) = 172 <0.000022>
[pid 487992] 13:13:40 close(7) = 0 <0.000026>
[pid 487992] 13:13:40 select(7, [6], [6], NULL, {tv_sec=1, tv_usec=0}) = 1 (out [6], left {tv_sec=0, tv_usec=999989}) <0.000032>
[pid 487992] 13:13:40 sendto(6, "delete user_session-0q77hvnob34bbh04t2859p2ro0\r\n", 48, MSG_NOSIGNAL, NULL, 0) = 48 <0.000125>
[pid 487992] 13:13:40 select(7, [6], [], NULL, {tv_sec=1, tv_usec=0}) = 1 (in [6], left {tv_sec=0, tv_usec=999998}) <0.000022>
[pid 487992] 13:13:40 recvfrom(6, "NOT_FOUND\r\n", 32768, 0, NULL, NULL) = 11 <0.000029>
[pid 487992] 13:13:40 select(5, [4], [4], NULL, {tv_sec=1, tv_usec=0}) = 1 (out [4], left {tv_sec=0, tv_usec=999996}) <0.000023>
[pid 487992] 13:13:40 sendto(4, "delete 0q77hvnob34bbh04t2859p2ro0\r\n", 35, MSG_NOSIGNAL, NULL, 0) = 35 <0.000074>
[pid 487992] 13:13:40 sendto(4, "delete 0q77hvnob34bbh04t2859p2ro0.lock\r\n", 40, MSG_NOSIGNAL, NULL, 0) = 40 <0.000067>
[pid 487992] 13:13:40 select(5, [4], [], NULL, {tv_sec=1, tv_usec=0}) = 1 (in [4], left {tv_sec=0, tv_usec=999998}) <0.000024>
[pid 487992] 13:13:40 recvfrom(4, "NOT_FOUND\r\nDELETED\r\n", 32768, 0, NULL, NULL) = 20 <0.000023>
[pid 487992] 13:13:40 close(4) = 0 <0.000084>
[pid 487992] 13:13:40 write(2, "zend_mm_heap corrupted\n", 23) = 23 <0.000018>
[pid 487992] 13:13:40 exit_group(1) = ?
[pid 487992] 13:13:40 +++ exited with 1 +++
[pid 487854] 13:13:40 <... poll resumed>) = 1 ([{fd=21, revents=POLLIN|POLLERR|POLLHUP}]) <0.335258>
[pid 485171] 13:13:40 <... epoll_wait resumed>0x55fbd8790a50, 1, 1000) = -1 EINTR (Interrupted system call) <0.401229>
[pid 487854] 13:13:40 read(21, <unfinished ...>
[pid 485171] 13:13:40 --- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=487992, si_uid=1001, si_status=1, si_utime=2, si_stime=3} ---
[pid 487854] 13:13:40 <... read resumed>0x7f7591d28a28, 8) = -1 ECONNRESET (Connection reset by peer) <0.000155>
[pid 485171] 13:13:40 <... rt_sigreturn resumed>) = -1 EINTR (Interrupted system call) <0.000149>
[pid 487854] 13:13:40 <... getpid resumed>) = 487760 <0.000142>
[pid 485171] 13:13:40 getpid( <unfinished ...>
[pid 487854] 13:13:40 write(9, "[Tue Jul 06 13:13:40.617431 2021] [proxy_fcgi:error] [pid 487760:tid 140142934406912] [client 86.252.93.61:49363] AH01067: Failed to read FastCGI header\n", 153 <unfinished ...>
[pid 485171] 13:13:40 <... getpid resumed>) = 485171 <0.000130>
[pid 487854] 13:13:40 <... write resumed>) = 153 <0.000193>
[pid 485171] 13:13:40 epoll_wait(8, <unfinished ...>
[pid 485171] 13:13:40 <... epoll_wait resumed>[{EPOLLIN, {u32=3627653888, u64=94540152806144}}], 1, 596) = 1 <0.000071>
[pid 487854] 13:13:40 <... getpid resumed>) = 487760 <0.000176>
[pid 487854] 13:13:40 write(9, "[Tue Jul 06 13:13:40.618309 2021] [proxy_fcgi:error] [pid 487760:tid 140142934406912] (104)Connection reset by peer: [client 86.252.93.61:49363] AH01075: Error dispatching request to : \n", 186 <unfinished ...>
[pid 485171] 13:13:40 <... read resumed>"C", 1) = 1 <0.000119>
[pid 487854] 13:13:40 <... write resumed>) = 186 <0.000226>
[pid 485171] 13:13:40 <... read resumed>0x7ffc8ee81f27, 1) = -1 EAGAIN (Resource temporarily unavailable) <0.000089>
[pid 487854] 13:13:40 <... close resumed>) = 0 <0.000158>