From c3fb7e1da9de554b8ffe3a2fac55900dc4b94ef9 Mon Sep 17 00:00:00 2001 From: Klaus Wuestefeld Date: Thu, 7 Dec 2023 15:43:53 -0300 Subject: [PATCH] Known issue added to debugging.md Multiple subsequent checkpoints don't work yet. --- debugging.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/debugging.md b/debugging.md index d507a0a..1731267 100644 --- a/debugging.md +++ b/debugging.md @@ -182,6 +182,10 @@ This is rather a sign that CRIU has insufficient privileges to write into `ns_la One trick that can be used in containers is to ensure that before the checkpoint PIDs are higher than anything needed for the restore, either writing `/proc/sys/kernel/ns_last_pid` or cycling dummy processes until `ns_last_pid` is higher than the required value (128 might be a good starting point). +## Multiple Checkpoints + +Doing a subsequent checkpoint on a process that has already been checkpointed and restored does not yet work. This is a [known issue](https://github.com/CRaC/docs/issues/21). + ## Further debugging of restore During restore CRIU writes its log into standard output with errors-only verbosity level (1). Debug-level (4) output can be enabled using VM option `-XX:CREngine=criuengine,--verbosity=4,--log-file=/path/to/log.txt` by passing these options to CRIU.