Tim Düsterhus

# Checking the old worker ``` master> @!26296 show sess all 0x55c03d4f9610: [26/Jul/2022:08:19:30.106266] id=0 proto=? flags=0x2, conn_retries=0, conn_exp= conn_et=0x000 srv_conn=(nil), pend_pos=(nil) waiting=0 epoch=0 frontend=tracker (id=4294967295 mode=peers), listener=? (id=0) backend=tracker (id=4294967295...

> So I'll fix these bugs. But, on your side, I guess you should fix your configuration to enable the SSL for your local peer server. Otherwise, the local resync...

And to understand this correctly: The old processes did not finish their soft stop, because the local peer resync failed? Just curious: Would `hard-stop-after` have fixed the issue?

> You can add the SSL config on the server line (the local one) or use a `default-server` directive. But the second solution will also enable the SSL on your...

@capflam I won't test the patches, as the workaround works for me, but one small comment: For 0001 in the error message: "local peer have" – the "have" should be...

> thanks to the attached patch file. @haproxyFred I believe you forgot the attachment :-)

> It was noticed right after 1.8 was released that issuing "show cache" on the CLI would show entries with negative expiration date. It should still be true on 1.9...

Related: #696

> Do you reproduce this problem for each reload? No, only for some: ``` Jul 25 13:11:07 *snip* systemd[1]: Reloading HAProxy Load Balancer. Jul 25 13:11:07 *snip* haproxy[1652444]: [NOTICE] (1652444)...

> So what's possible is that the new process couldn't connect to the previous one, the error you are seeing are emitted when trying to do a send() or a...