serverfault.com

We run an application on Google App Engine Flexible Environment with two services: Service default: machine e2-custom-4-8704, min_num_instances: 1, disk: pd-standard Service aux: machine e2-custom-2-6656, min_num_instances: 1, disk: pd-standard Region: southamerica-east1 Runtime: custom (Docker) Network: Shared VPC between two GCP projects What happened On 2026-03-23, between 20:18 and 20:21 UTC (17:18–17:21 Brasília time), our application was completely unavailable for approximately 3 minutes. Users saw: Error: Server Error The server encountered a temporary error and could not complete your request. Please try again in 30 seconds. This was NOT a manual deploy. Cloud Audit Logs confirm the entire process was triggered automatically by GCP's own service account (GCP's internal App Engine service account) as part of the weekly OS rotation. The same behavior occurred again the following week, confirming it is reproducible. Audit log evidence The sequence from Cloud Au

Scenario: Customer with hybrid Exchange environment; all mailboxes are in Exchange Online. Latest Exchange version (SE). Exchange was originally deployed in a resource forest topology, thus each mailbox existed in the resource forest and was linked to a user account in the main forest. Entra Connect syncs both forests and merges the Exchange attributes in the resulting user objects. All mailboxes have been migrated to Exchange Online. Now, when the customer needs to create a new mailbox, they use the following process: The mailbox is created as a linked mailbox in the on-premises Exchange server. The mailbox is them migrated to Exchange Online. From that point on, the mailbox is treated by Exchange as a remote mailbox. It would be a lot easier to simply create a remote mailbox, but it looks like this is not possible: the

I am the owner of a real-time utility website, spectrumoutage.org, which tracks internet service outages. I am currently facing a persistent "Ad Serving Limit" from Google AdSense due to "Invalid Traffic" concerns, and I am looking for help analyzing what on the site might be triggering this. The Situation: The site provides live updates on ISP outages. Because of the nature of the content, users tend to refresh the pages frequently to see if their service has been restored. What I've Checked: Traffic Sources: According to my analytics, the majority of traffic is organic search or direct. I am not using any paid traffic or bots. Ad Placement: I am using standard AdSense Auto-Ads. Content: The content is utility-driven and unique to regional outages. My Questions for the Community: Could the high frequency of user refr

I am currently going somewhat crazy debugging an issue regarding my OU and the way GPO's are applied. I have noticed, that there are invisible undocumented rules being applied to any servers within this built in OU "Domain Controllers". Servers behave differently when they are in that OU as oppose to a different one even when the applied gpo are identical. (i turn off inheritance and link the same ones to both OU). The output of gpresult confirms that the policies are the same. I obviously do not want to have to put non dc servers into my dc OU. The only reference or documentation i can find to this issue is this https://community.spiceworks.com/t/whats-special-about-the-domain-controllers-ou/930162 which is sadly not very helpful and points to a dead link for an article that explains further. Any information you

If LACP declaration (server side) will still send traffic when one of the NICs fails, is there any advantage to declaring a bond as one primary one slave? Other than having the backup on a financially cheaper route I cannot see any positive to it. Is there any valid technical argument in favour? I am trying to steer our automation toward LACP as a install default rather than active_backup as it is now.

Kubernetes supports user namespaces, and apparently supports id-mapped mounts. Tools like podman allow for using id-mapped mounts via the idmap option to --mount, but there doesn't appear to be anything in the kubernetes docs about how to use id-mapped mounts. What are the options that are required to be set in the PV/PVC/Pod config for performing an id-mapped mount?

I have an active-backup Linux network bond with the following setup. # bond0 modprobe bonding # add_master ip link add dev bond0 type bond # setup_master ip link set dev bond0 type bond mode active-backup miimon 100 ip link set dev bond0 type bond primary_reselect failure # enslave_slaves ip link set dev wlan0 master bond0 ip link set dev wlan1 master bond0 ip link set dev bond0 type bond primary wlan1 The wlan0 interface is connected to a Linux AP device which is bridged over ethernet to a consumer router. The wlan1 interface is directly connected to the same consumer router. Then the following occurs: Primary goes down Backup becomes active Primary recovers and becomes active again The following then happens: The Linux device connected on the other end ofwlan0 CAN ssh into it and make any connections. However, on ALL other

I'm trying to create an Azure OpenAI resource on an Azure for Students subscription and validation fails at the Review + submit step with: Resource 'LongLaMP-student' was disallowed by Azure: This policy maintains a set of best available regions where your subscription can deploy resources. The objective of this policy is to ensure that your subscription has full access to Azure services with optimal performance. Should you need additional or different regions, contact support. (Code: RequestDisallowedByAzure, Target: LongLaMP-student) Deployment parameters: Subscription: Azure for Students Resource group: LongLaMP-student (created successfully in the same subscription) Region: East US 2 Name: LongLaMP-student Pricing tier: Standard S0 Network: All networks

This error seems to also occur on other users' clusters, but I haven't found a working remedy for it so far: On my Ceph cluster (4 nodes; Ceph 19.2.3 Squid (stable) on Ubuntu 24.04.4 LTS (aarch64)), I've managed to add all hosts: mixtile@blade3n1:~$ sudo ceph orch host ls HOST ADDR LABELS STATUS blade3n1 10.20.0.11 _admin blade3n2 10.20.0.12 blade3n3 10.20.0.13 blade3n4 10.20.0.14 4 hosts in cluster …but trying to create the necessary OSDs by using ceph orch apply osd --all-available-devices leaves me with such an error message in the cephadm log: 2026-04-21 18:50:52,523 ffffbe34e720 INFO /usr/bin/docker: stderr Running command: /usr/bin/ceph-osd --cluster ceph --osd-objectstore bluestore --mkfs -i 2 --monmap /var/lib/ceph/osd/ceph-2/activate.monmap --keyfile - --osdspec-affinity all-available-devices --osd-data /var/lib/ceph/osd/ceph-2/ --osd-uuid 3ad68f34-9ca7-4430-9870-0f563c64f631 --setu

I have a system (Ubuntu 24.04) with a single NIC. The server gets an IP via DHCP from an external system. This server is running virtual machines via KVM. I would like to create a network bridge that will allow the VMs on the system to also retrieve their IPs from the same external system that's the source for the server's IP, instead of getting their IPs from the DHCP server on the local system. So, upon booting, the server would get an IP (e.g. 86.16.17.4) from the remote system, and then the VMs would autostart and get their own IPs (e.g. 86.16.17.5, 86.16.17.6) from the same remote system. Can this be achieved with only a single NIC? If so, how? Everything I've found so far assumes you're either working with multiple NICs, or you have one NIC and all of the VMs are getting their IPs from the DHCP server on the local system.