ai-agents-retry-fallback-ollama-llm-calls-intermittently-fail-with-http-400-c8ae2855

THEN Wrap the chain (e.g., `rag_chain = prompt | llm | StrOutputParser()`) with `.with_retry()` or use fallback chains via `.with_fallbacks()` to automatically retry on transient failures. For example: `rag_chain.with_retry()`.

from langchain.prompts import PromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_community.chat_models import ChatOllama

llm = ChatOllama(model='llama3', temperature=0)
prompt = PromptTemplate.from_template('...')
rag_chain = prompt | llm | StrOutputParser()
rag_chain = rag_chain.with_retry()  # Add retry for transient Ollama errors
question = 'agent memory'
docs = retriever.invoke(question)
generation = rag_chain.invoke({'question': question, 'context': docs})