Bug 2388201

Created attachment 2103444 [details]
File: anaconda-tb

Created attachment 2103445 [details]
File: backtrace

Created attachment 2103446 [details]
File: os_info

Created attachment 2103447 [details]
File: environ

Created attachment 2103448 [details]
File: syslog

Created attachment 2103449 [details]
File: description

Created attachment 2103450 [details]
File: storage.log

Created attachment 2103451 [details]
File: packaging.log

Here is the patch, to fix this crash, generated by cursor, I'm not good at c, but it looks like a valid fix:

-       if (stat(fname, &stb))
-               goto out;
-       if (!S_ISDIR(stb.st_mode))
-               goto out;
-       strncpy(mdi->sys_name, devnm, sizeof(mdi->sys_name) - 1);
+       /* 
+        * In some cases (such as during device assembly), the kernel may need
+        * some time to create sysfs entries. We add a retry mechanism to handle
+        * this timing issue.
+        */
+       while (retry_count <= max_retries) {
+               if (stat(fname, &stb) == 0) {
+                       if (S_ISDIR(stb.st_mode)) {
+                               strncpy(mdi->sys_name, devnm, sizeof(mdi->sys_name) - 1);
+                               retval = 0;
+                               break;
+                       } else {
+                               /* Path exists but is not a directory, this is an error */
+                               goto out;
+                       }
+               }
+               
+               /* If first attempt fails, we record debug information */
+               if (retry_count == 0) {
+                       /* Don't print error here as this may be temporary during normal operation */
+               }
+               
+               /* If maximum retries reached, exit */
+               if (retry_count == max_retries)
+                       break;
+               
+               /* Wait for a while before retrying */
+               usleep(retry_delay_ms * 1000);
+               retry_count++;
+       }

-       retval = 0;
 out:
        return retval;
 }
--

Created attachment 2103452 [details]
File: program.log

Created attachment 2103453 [details]
File: nmcli_dev_list

Created attachment 2103454 [details]
File: anaconda.log

Created attachment 2103455 [details]
File: lorax-packages.log

Created attachment 2103456 [details]
File: lsblk_output

Created attachment 2103458 [details]
File: dbus.log

Created attachment 2103459 [details]
File: packaging.log

Created attachment 2103732 [details]
screenshot

Forgot to mention the reproducor:

1)boot  Fedora-Server-dvd-x86_64-43-20250812.n.1.iso on a system with an existing raid, 
2)choose "blivet-gui" on installation destination page,  delete that raid device and create a new one ,
3)the error will happen immediately after storage configure progress starts.

As shown in the screenshot I just attached, this same error message will pop up when I tried to start a mdraid device on anaconda-webui

Proposed as a Blocker for 43-beta by Fedora user lnie using the blocker tracking app because:

 seems violates:
https://fedoraproject.org/wiki/Fedora_43_Beta_Release_Criteria#Custom_partitioning

+6 in https://pagure.io/fedora-qa/blocker-review/issue/1872 , marking accepted (though it'd be nice if anyone else can confirm this; I'll see if I have time to try in a bit).

Hi Inie

Which file is it that the patch you mentioned in comment9? (what's the diff output)

Thanks
Xiao

Created attachment 2104852 [details]
diff

hi,

I uploaded the diff file just now.

Regards,
Lili

Hi Inie

This problem is same with https://bugzilla.redhat.com/show_bug.cgi?id=2388480. f43 already has a update to fix this.

The original bug is https://bugzilla.redhat.com/show_bug.cgi?id=2385871 which is aginst f44

So this should be fixed in recent composes? lnie, can you re-test with the most recent F43 compose? Thanks.

> lnie, can you re-test with the most recent F43 compose? Thanks.

Sure:)

check with Fedora-Server-dvd-x86_64-43-20250826.n.1.iso which contains mdadm-4.3-9.fc43, the issue is gone.

Thanks for confirming.